digital-forensics-labs/mobile-pii-discovery-agent
1
0
Fork 0
mirror of https://github.com/frankwxu/mobile-pii-discovery-agent.git synced 2026-02-20 13:40:41 +00:00
Code Issues Packages Projects Releases Wiki Activity

add other model results

Browse Source
This commit is contained in:
Frank Xu
2026-02-11 22:29:04 -05:00
parent ef30c3de38
commit a80dabe005
88 changed files with 589 additions and 270 deletions
Download Patch File Download Diff File Expand all files Collapse all files

108
RQs/RQ0/RQ0_1_results_normalization.ipynb
View File

@@ -2,7 +2,7 @@
"cells": [ "cells": [
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": null, "execution_count": 3,
"id": "c2d824a6", "id": "c2d824a6",
"metadata": {}, "metadata": {},
"outputs": [ "outputs": [
@@ -10,58 +10,58 @@
"name": "stdout", "name": "stdout",
"output_type": "stream", "output_type": "stream",
"text": [ "text": [
"PII_A1_commerce_20260131T203324Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_A1_commerce_20260131T203324Z.jsonl\n", "PII_A1_commerce_20260211T022802Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_A1_commerce_20260211T022802Z.jsonl\n",
"PII_A1_msgstore_20260131T203502Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_A1_msgstore_20260131T203502Z.jsonl\n", "PII_A1_msgstore_20260211T024003Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_A1_msgstore_20260211T024003Z.jsonl\n",
"PII_A1_wa_20260131T203943Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_A1_wa_20260131T203943Z.jsonl\n", "PII_A1_wa_20260211T024706Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_A1_wa_20260211T024706Z.jsonl\n",
"PII_A2_core_20260131T204055Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_A2_core_20260131T204055Z.jsonl\n", "PII_A2_core_20260211T023156Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_A2_core_20260211T023156Z.jsonl\n",
"PII_A2_journal_20260131T204142Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_A2_journal_20260131T204142Z.jsonl\n", "PII_A2_journal_20260211T023216Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_A2_journal_20260211T023216Z.jsonl\n",
"PII_A2_main_20260131T204345Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_A2_main_20260131T204345Z.jsonl\n", "PII_A2_main_20260211T023611Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_A2_main_20260211T023611Z.jsonl\n",
"PII_A3_account1cache4_20260131T204410Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_A3_account1cache4_20260131T204410Z.jsonl\n", "PII_A3_account1cache4_20260211T023627Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_A3_account1cache4_20260211T023627Z.jsonl\n",
"PII_A3_account2cache4_20260131T204617Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_A3_account2cache4_20260131T204617Z.jsonl\n", "PII_A3_account2cache4_20260211T023643Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_A3_account2cache4_20260211T023643Z.jsonl\n",
"PII_A3_account3cache4_20260131T204642Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_A3_account3cache4_20260131T204642Z.jsonl\n", "PII_A3_account3cache4_20260211T023659Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_A3_account3cache4_20260211T023659Z.jsonl\n",
"PII_A4_gmm_myplaces_20260131T204709Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_A4_gmm_myplaces_20260131T204709Z.jsonl\n", "PII_A4_gmm_myplaces_20260211T025539Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_A4_gmm_myplaces_20260211T025539Z.jsonl\n",
"PII_A4_gmm_storage_20260131T204738Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_A4_gmm_storage_20260131T204738Z.jsonl\n", "PII_A4_gmm_storage_20260211T025558Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_A4_gmm_storage_20260211T025558Z.jsonl\n",
"PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14_20260131T204833Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14_20260131T204833Z.jsonl\n", "PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14_20260211T025625Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14_20260211T025625Z.jsonl\n",
"PII_A5_SBrowser2_20260131T205010Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_A5_SBrowser2_20260131T205010Z.jsonl\n", "PII_A5_SBrowser2_20260211T025812Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_A5_SBrowser2_20260211T025812Z.jsonl\n",
"PII_A5_SBrowser_20260131T204925Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_A5_SBrowser_20260131T204925Z.jsonl\n", "PII_A5_SBrowser_20260211T025741Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_A5_SBrowser_20260211T025741Z.jsonl\n",
"PII_A5_searchengine_20260131T205101Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_A5_searchengine_20260131T205101Z.jsonl\n", "PII_A5_searchengine_20260211T025835Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_A5_searchengine_20260211T025835Z.jsonl\n",
"PII_I1_CallHistory_20260131T205155Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_I1_CallHistory_20260131T205155Z.jsonl\n", "PII_I1_CallHistory_20260211T025226Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_I1_CallHistory_20260211T025226Z.jsonl\n",
"PII_I1_ChatStorage_20260131T205309Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_I1_ChatStorage_20260131T205309Z.jsonl\n", "PII_I1_ChatStorage_20260211T025417Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_I1_ChatStorage_20260211T025417Z.jsonl\n",
"PII_I1_ContactsV2_20260131T210034Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_I1_ContactsV2_20260131T210034Z.jsonl\n", "PII_I1_ContactsV2_20260211T030947Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_I1_ContactsV2_20260211T030947Z.jsonl\n",
"PII_I2_AddressBook_20260131T210607Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_I2_AddressBook_20260131T210607Z.jsonl\n", "PII_I2_AddressBook_20260211T030143Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_I2_AddressBook_20260211T030143Z.jsonl\n",
"PII_I2_AddressBookImages_20260131T210641Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_I2_AddressBookImages_20260131T210641Z.jsonl\n", "PII_I2_AddressBookImages_20260211T030157Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_I2_AddressBookImages_20260211T030157Z.jsonl\n",
"PII_I3_sms_20260131T210735Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_I3_sms_20260131T210735Z.jsonl\n", "PII_I3_sms_20260211T030251Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_I3_sms_20260211T030251Z.jsonl\n",
"PII_I4_CloudTabs_20260131T210821Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_I4_CloudTabs_20260131T210821Z.jsonl\n", "PII_I4_CloudTabs_20260211T030321Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_I4_CloudTabs_20260211T030321Z.jsonl\n",
"PII_I4_History_20260131T210912Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_I4_History_20260131T210912Z.jsonl\n", "PII_I4_History_20260211T030358Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_I4_History_20260211T030358Z.jsonl\n",
"PII_I5_Calendar_20260131T211008Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_I5_Calendar_20260131T211008Z.jsonl\n", "PII_I5_Calendar_20260211T030514Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_I5_Calendar_20260211T030514Z.jsonl\n",
"PII_I5_Extras_20260131T211054Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\\PII_I5_Extras_20260131T211054Z.jsonl\n", "PII_I5_Extras_20260211T030537Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\\PII_I5_Extras_20260211T030537Z.jsonl\n",
"Done. Files: 25, Records: 125\n", "Done. Files: 25, Records: 125\n",
"Output folder: I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\db_level\n", "Output folder: I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\db_level\n",
"PII_A1_commerce_20260127T175911Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A1_commerce_20260127T175911Z.jsonl\n", "PII_A1_commerce.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A1_commerce.jsonl\n",
"PII_A1_msgstore_20260127T180043Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A1_msgstore_20260127T180043Z.jsonl\n", "PII_A1_msgstore.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A1_msgstore.jsonl\n",
"PII_A1_wa_20260127T180213Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A1_wa_20260127T180213Z.jsonl\n", "PII_A1_wa.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A1_wa.jsonl\n",
"PII_A2_core_20260127T180339Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A2_core_20260127T180339Z.jsonl\n", "PII_A2_core.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A2_core.jsonl\n",
"PII_A2_journal_20260127T180440Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A2_journal_20260127T180440Z.jsonl\n", "PII_A2_journal.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A2_journal.jsonl\n",
"PII_A2_main_20260127T180710Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A2_main_20260127T180710Z.jsonl\n", "PII_A2_main.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A2_main.jsonl\n",
"PII_A3_account1cache4_20260127T180745Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A3_account1cache4_20260127T180745Z.jsonl\n", "PII_A3_account1cache4.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A3_account1cache4.jsonl\n",
"PII_A3_account2cache4_20260127T180821Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A3_account2cache4_20260127T180821Z.jsonl\n", "PII_A3_account2cache4.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A3_account2cache4.jsonl\n",
"PII_A3_account3cache4_20260127T180857Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A3_account3cache4_20260127T180857Z.jsonl\n", "PII_A3_account3cache4.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A3_account3cache4.jsonl\n",
"PII_A4_gmm_myplaces_20260127T180935Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A4_gmm_myplaces_20260127T180935Z.jsonl\n", "PII_A4_gmm_myplaces.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A4_gmm_myplaces.jsonl\n",
"PII_A4_gmm_storage_20260127T181014Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A4_gmm_storage_20260127T181014Z.jsonl\n", "PII_A4_gmm_storage.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A4_gmm_storage.jsonl\n",
"PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14_20260127T181121Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14_20260127T181121Z.jsonl\n", "PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14.jsonl\n",
"PII_A5_SBrowser2_20260127T181345Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A5_SBrowser2_20260127T181345Z.jsonl\n", "PII_A5_SBrowser.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A5_SBrowser.jsonl\n",
"PII_A5_SBrowser_20260127T181239Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A5_SBrowser_20260127T181239Z.jsonl\n", "PII_A5_SBrowser2.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A5_SBrowser2.jsonl\n",
"PII_A5_searchengine_20260127T181446Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A5_searchengine_20260127T181446Z.jsonl\n", "PII_A5_searchengine.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_A5_searchengine.jsonl\n",
"PII_I1_CallHistory_20260127T181557Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I1_CallHistory_20260127T181557Z.jsonl\n", "PII_I1_CallHistory.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I1_CallHistory.jsonl\n",
"PII_I1_ChatStorage_20260127T181731Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I1_ChatStorage_20260127T181731Z.jsonl\n", "PII_I1_ChatStorage.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I1_ChatStorage.jsonl\n",
"PII_I1_ContactsV2_20260127T182906Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I1_ContactsV2_20260127T182906Z.jsonl\n", "PII_I1_ContactsV2.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I1_ContactsV2.jsonl\n",
"PII_I2_AddressBook_20260127T183457Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I2_AddressBook_20260127T183457Z.jsonl\n", "PII_I2_AddressBook.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I2_AddressBook.jsonl\n",
"PII_I2_AddressBookImages_20260127T183526Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I2_AddressBookImages_20260127T183526Z.jsonl\n", "PII_I2_AddressBookImages.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I2_AddressBookImages.jsonl\n",
"PII_I3_sms_20260127T183606Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I3_sms_20260127T183606Z.jsonl\n", "PII_I3_sms.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I3_sms.jsonl\n",
"PII_I4_CloudTabs_20260127T183643Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I4_CloudTabs_20260127T183643Z.jsonl\n", "PII_I4_CloudTabs.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I4_CloudTabs.jsonl\n",
"PII_I4_History_20260127T183727Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I4_History_20260127T183727Z.jsonl\n", "PII_I4_History.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I4_History.jsonl\n",
"PII_I5_Calendar_20260127T183815Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I5_Calendar_20260127T183815Z.jsonl\n", "PII_I5_Calendar.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I5_Calendar.jsonl\n",
"PII_I5_Extras_20260127T183857Z.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I5_Extras_20260127T183857Z.jsonl\n", "PII_I5_Extras.jsonl: 5 records -> I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\\PII_I5_Extras.jsonl\n",
"Done. Files: 25, Records: 125\n", "Done. Files: 25, Records: 125\n",
"Output folder: I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\n" "Output folder: I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\db_level\n"
] ]
@@ -139,8 +139,8 @@
"\n", "\n",
" from stats_utils import normalize_and_slim_record\n", " from stats_utils import normalize_and_slim_record\n",
"\n", "\n",
" IN_DIR = Path(r\"..\\..\\model_PII_results\\gpt4o\")\n", " IN_DIR = Path(r\"..\\..\\model_PII_results\\GPT-5.1\")\n",
" RESULTS_DIR = Path(r\"normalized_PII_results\\gpt4o\\db_level\")\n", " RESULTS_DIR = Path(r\"normalized_PII_results\\GPT-5.1\\db_level\")\n",
" OUT_DIR = STATS_DIR/ RESULTS_DIR \n", " OUT_DIR = STATS_DIR/ RESULTS_DIR \n",
"\n", "\n",
" normalize_jsonl_folder(IN_DIR, OUT_DIR, normalize_and_slim_record, delete_out_dir_first=True)\n", " normalize_jsonl_folder(IN_DIR, OUT_DIR, normalize_and_slim_record, delete_out_dir_first=True)\n",

8
RQs/RQ0/RQ0_2_app_level_aggregation.ipynb
View File

@@ -2,7 +2,7 @@
"cells": [ "cells": [
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": null, "execution_count": 2,
"id": "234eed3f", "id": "234eed3f",
"metadata": {}, "metadata": {},
"outputs": [ "outputs": [
@@ -10,7 +10,7 @@
"name": "stdout", "name": "stdout",
"output_type": "stream", "output_type": "stream",
"text": [ "text": [
"Wrote: I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\app_level\\app_level.jsonl\n", "Wrote: I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\app_level\\app_level.jsonl\n",
"Wrote: I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\app_level\\app_level.jsonl\n" "Wrote: I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\app_level\\app_level.jsonl\n"
] ]
} }
@@ -182,8 +182,8 @@
"\n", "\n",
"if __name__ == \"__main__\": \n", "if __name__ == \"__main__\": \n",
" # --- Aggregate GPT-4o results ---\n", " # --- Aggregate GPT-4o results ---\n",
" IN_DIR = Path(r\"..\\normalized_PII_results\\gpt4o\\db_level\")\n", " IN_DIR = Path(r\"..\\normalized_PII_results\\GPT-5.1\\db_level\")\n",
" OUT_DIR = Path(r\"..\\normalized_PII_results\\gpt4o\\app_level\")\n", " OUT_DIR = Path(r\"..\\normalized_PII_results\\GPT-5.1\\app_level\")\n",
" OUT_DIR.mkdir(parents=True, exist_ok=True)\n", " OUT_DIR.mkdir(parents=True, exist_ok=True)\n",
"\n", "\n",
" out_path = OUT_DIR / \"app_level.jsonl\"\n", " out_path = OUT_DIR / \"app_level.jsonl\"\n",

6
RQs/RQ0/RQ0_3_corpus_level_aggreation.ipynb
View File

@@ -10,7 +10,7 @@
"name": "stdout", "name": "stdout",
"output_type": "stream", "output_type": "stream",
"text": [ "text": [
"Wrote: I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\gpt4o\\corpus_level\\corpus_level.jsonl\n", "Wrote: I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\GPT-5.1\\corpus_level\\corpus_level.jsonl\n",
"Wrote: I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\corpus_level\\corpus_level.jsonl\n" "Wrote: I:\\project2026\\llmagent\\RQs\\normalized_PII_results\\ground_truth\\corpus_level\\corpus_level.jsonl\n"
] ]
} }
@@ -146,8 +146,8 @@
"\n", "\n",
"if __name__ == \"__main__\":\n", "if __name__ == \"__main__\":\n",
" # Aggregate GPT-4o results\n", " # Aggregate GPT-4o results\n",
" IN_DIR = Path(r\"..\\normalized_PII_results\\gpt4o\\db_level\")\n", " IN_DIR = Path(r\"..\\normalized_PII_results\\GPT-5.1\\db_level\")\n",
" OUT_DIR = Path(r\"..\\normalized_PII_results\\gpt4o\\corpus_level\\corpus_level.jsonl\")\n", " OUT_DIR = Path(r\"..\\normalized_PII_results\\GPT-5.1\\corpus_level\\corpus_level.jsonl\")\n",
" out = aggregate_jsonl_folder_corpus_level(\n", " out = aggregate_jsonl_folder_corpus_level(\n",
" IN_DIR,\n", " IN_DIR,\n",
" OUT_DIR\n", " OUT_DIR\n",

36
RQs/RQ1/RQ1_t4.ipynb
View File

@@ -2,7 +2,7 @@
"cells": [ "cells": [
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 2, "execution_count": 1,
"id": "e15e3ffe", "id": "e15e3ffe",
"metadata": {}, "metadata": {},
"outputs": [ "outputs": [
@@ -13,29 +13,29 @@
"ID Application Database Email Phone UserName PersonName PostalAddr Total\n", "ID Application Database Email Phone UserName PersonName PostalAddr Total\n",
"-------------------------------------------------------------------------------------------------\n", "-------------------------------------------------------------------------------------------------\n",
"A1 WhatsApp commerce.db 0 0 0 0 0 0\n", "A1 WhatsApp commerce.db 0 0 0 0 0 0\n",
"A1 WhatsApp msgstore.db 0 7 5 15 0 27\n", "A1 WhatsApp msgstore.db 0 0 3 26 0 29\n",
"A1 WhatsApp wa.db 0 16 680 10 0 706\n", "A1 WhatsApp wa.db 0 408 0 129 0 537\n",
"A2 Snapchat core.db 0 1 5 1 0 7\n", "A2 Snapchat core.db 1 1 1 0 0 3\n",
"A2 Snapchat journal.db 0 0 0 0 0 0\n", "A2 Snapchat journal.db 0 0 0 0 0 0\n",
"A2 Snapchat main.db 1 13 0 11 0 25\n", "A2 Snapchat main.db 1 13 1229 252 0 1495\n",
"A3 Telegram account1cache4.db 0 0 0 0 0 0\n", "A3 Telegram account1cache4.db 0 0 0 0 0 0\n",
"A3 Telegram account2cache4.db 0 0 0 0 0 0\n", "A3 Telegram account2cache4.db 0 0 0 0 0 0\n",
"A3 Telegram account3cache4.db 0 0 0 0 0 0\n", "A3 Telegram account3cache4.db 0 0 0 0 0 0\n",
"A4 Google Maps gmm_myplaces.db 0 0 0 0 0 0\n", "A4 Google Maps gmm_myplaces.db 0 0 0 0 0 0\n",
"A4 Google Maps gmm_storage.db 0 0 0 0 0 0\n", "A4 Google Maps gmm_storage.db 0 0 0 0 0 0\n",
"A4 Google Maps peopleCache_sh....db 2 0 2 0 0 4\n", "A4 Google Maps peopleCache_sh....db 2 1 2 2 0 7\n",
"A5 Samsung Internet SBrowser.db 1 0 5 0 0 6\n", "A5 Samsung Internet SBrowser.db 1 0 0 2 0 3\n",
"A5 Samsung Internet SBrowser2.db 0 0 0 0 0 0\n", "A5 Samsung Internet SBrowser2.db 0 0 0 1 0 1\n",
"A5 Samsung Internet searchengine.db 0 0 17 0 0 17\n", "A5 Samsung Internet searchengine.db 0 0 0 0 0 0\n",
"I1 WhatsApp CallHistory.sqlite 0 0 10 0 0 10\n", "I1 WhatsApp CallHistory.sqlite 0 3 0 0 0 3\n",
"I1 WhatsApp ChatStorage.sqlite 0 0 0 14 2 16\n", "I1 WhatsApp ChatStorage.sqlite 0 63 0 66 2 131\n",
"I1 WhatsApp ContactsV2.sqlite 0 655 0 0 0 655\n", "I1 WhatsApp ContactsV2.sqlite 0 0 0 1350 0 1350\n",
"I2 Contacts AddressBook.sqlitedb 6 548 1 748 0 1303\n", "I2 Contacts AddressBook.sqlitedb 6 716 0 733 0 1455\n",
"I2 Contacts AddressB....sqlitedb 0 0 0 0 0 0\n", "I2 Contacts AddressB....sqlitedb 0 0 0 0 0 0\n",
"I3 Apple Messages sms.db 1 0 0 10 0 11\n", "I3 Apple Messages sms.db 7 0 0 14 0 21\n",
"I4 Safari CloudTabs.db 0 0 25 0 0 25\n", "I4 Safari CloudTabs.db 0 0 0 0 0 0\n",
"I4 Safari History.db 0 0 4 0 0 4\n", "I4 Safari History.db 0 0 0 0 0 0\n",
"I5 Calendar Calendar.sqlitedb 1 0 0 0 0 1\n", "I5 Calendar Calendar.sqlitedb 1 0 0 1 0 2\n",
"I5 Calendar Extras.db 0 0 0 0 0 0\n", "I5 Calendar Extras.db 0 0 0 0 0 0\n",
"\n", "\n",
"Wrote LaTeX: I:\\project2026\\llmagent\\RQs\\RQ1\\RQ1_t4.tex\n", "Wrote LaTeX: I:\\project2026\\llmagent\\RQs\\RQ1\\RQ1_t4.tex\n",
@@ -306,7 +306,7 @@
"\n", "\n",
"\n", "\n",
"def main() -> None:\n", "def main() -> None:\n",
" in_dir = Path(r\"..\\normalized_PII_results\\gpt4o\\db_level\")\n", " in_dir = Path(r\"..\\normalized_PII_results\\GPT-5.1\\db_level\")\n",
" out_tex = Path(\"RQ1_t4.tex\")\n", " out_tex = Path(\"RQ1_t4.tex\")\n",
" out_csv = Path(\"RQ1_t4_plain.csv\")\n", " out_csv = Path(\"RQ1_t4_plain.csv\")\n",
" generate_db_level_pii_tables(in_dir, out_tex, out_csv)\n", " generate_db_level_pii_tables(in_dir, out_tex, out_csv)\n",

32
RQs/RQ1/RQ1_t4.tex
View File

@@ -9,15 +9,15 @@
\hline \hline
\multirow{3}{*}{A1} & \multirow{3}{*}{WhatsApp} & commerce.db & 0 & 0 & 0 & 0 & 0 & \textbf{0} \\ \multirow{3}{*}{A1} & \multirow{3}{*}{WhatsApp} & commerce.db & 0 & 0 & 0 & 0 & 0 & \textbf{0} \\
\cline{3-9} \cline{3-9}
& & msgstore.db & 0 & 7 & 5 & 15 & 0 & \textbf{27} \\ & & msgstore.db & 0 & 0 & 3 & 26 & 0 & \textbf{29} \\
\cline{3-9} \cline{3-9}
& & wa.db & 0 & 16 & 680 & 10 & 0 & \textbf{706} \\ & & wa.db & 0 & 408 & 0 & 129 & 0 & \textbf{537} \\
\hline \hline
\multirow{3}{*}{A2} & \multirow{3}{*}{Snapchat} & core.db & 0 & 1 & 5 & 1 & 0 & \textbf{7} \\ \multirow{3}{*}{A2} & \multirow{3}{*}{Snapchat} & core.db & 1 & 1 & 1 & 0 & 0 & \textbf{3} \\
\cline{3-9} \cline{3-9}
& & journal.db & 0 & 0 & 0 & 0 & 0 & \textbf{0} \\ & & journal.db & 0 & 0 & 0 & 0 & 0 & \textbf{0} \\
\cline{3-9} \cline{3-9}
& & main.db & 1 & 13 & 0 & 11 & 0 & \textbf{25} \\ & & main.db & 1 & 13 & 1229 & 252 & 0 & \textbf{1495} \\
\hline \hline
\multirow{3}{*}{A3} & \multirow{3}{*}{Telegram} & account1cache4.db & 0 & 0 & 0 & 0 & 0 & \textbf{0} \\ \multirow{3}{*}{A3} & \multirow{3}{*}{Telegram} & account1cache4.db & 0 & 0 & 0 & 0 & 0 & \textbf{0} \\
\cline{3-9} \cline{3-9}
@@ -29,31 +29,31 @@
\cline{3-9} \cline{3-9}
& & gmm\_storage.db & 0 & 0 & 0 & 0 & 0 & \textbf{0} \\ & & gmm\_storage.db & 0 & 0 & 0 & 0 & 0 & \textbf{0} \\
\cline{3-9} \cline{3-9}
& & peopleCache\_sh....db & 2 & 0 & 2 & 0 & 0 & \textbf{4} \\ & & peopleCache\_sh....db & 2 & 1 & 2 & 2 & 0 & \textbf{7} \\
\hline \hline
\multirow{3}{*}{A5} & \multirow{3}{*}{\begin{tabular}[c]{@{}l@{}}Samsung \\Internet\end{tabular}} & SBrowser.db & 1 & 0 & 5 & 0 & 0 & \textbf{6} \\ \multirow{3}{*}{A5} & \multirow{3}{*}{\begin{tabular}[c]{@{}l@{}}Samsung \\Internet\end{tabular}} & SBrowser.db & 1 & 0 & 0 & 2 & 0 & \textbf{3} \\
\cline{3-9} \cline{3-9}
& & SBrowser2.db & 0 & 0 & 0 & 0 & 0 & \textbf{0} \\ & & SBrowser2.db & 0 & 0 & 0 & 1 & 0 & \textbf{1} \\
\cline{3-9} \cline{3-9}
& & searchengine.db & 0 & 0 & 17 & 0 & 0 & \textbf{17} \\ & & searchengine.db & 0 & 0 & 0 & 0 & 0 & \textbf{0} \\
\hline \hline
\multirow{3}{*}{I1} & \multirow{3}{*}{WhatsApp} & CallHistory.sqlite & 0 & 0 & 10 & 0 & 0 & \textbf{10} \\ \multirow{3}{*}{I1} & \multirow{3}{*}{WhatsApp} & CallHistory.sqlite & 0 & 3 & 0 & 0 & 0 & \textbf{3} \\
\cline{3-9} \cline{3-9}
& & ChatStorage.sqlite & 0 & 0 & 0 & 14 & 2 & \textbf{16} \\ & & ChatStorage.sqlite & 0 & 63 & 0 & 66 & 2 & \textbf{131} \\
\cline{3-9} \cline{3-9}
& & ContactsV2.sqlite & 0 & 655 & 0 & 0 & 0 & \textbf{655} \\ & & ContactsV2.sqlite & 0 & 0 & 0 & 1350 & 0 & \textbf{1350} \\
\hline \hline
\multirow{2}{*}{I2} & \multirow{2}{*}{Contacts} & AddressBook.sqlitedb & 6 & 548 & 1 & 748 & 0 & \textbf{1303} \\ \multirow{2}{*}{I2} & \multirow{2}{*}{Contacts} & AddressBook.sqlitedb & 6 & 716 & 0 & 733 & 0 & \textbf{1455} \\
\cline{3-9} \cline{3-9}
& & AddressB....sqlitedb & 0 & 0 & 0 & 0 & 0 & \textbf{0} \\ & & AddressB....sqlitedb & 0 & 0 & 0 & 0 & 0 & \textbf{0} \\
\hline \hline
\multirow{1}{*}{I3} & \multirow{1}{*}{\begin{tabular}[c]{@{}l@{}}Apple \\Messages\end{tabular}} & sms.db & 1 & 0 & 0 & 10 & 0 & \textbf{11} \\ \multirow{1}{*}{I3} & \multirow{1}{*}{\begin{tabular}[c]{@{}l@{}}Apple \\Messages\end{tabular}} & sms.db & 7 & 0 & 0 & 14 & 0 & \textbf{21} \\
\hline \hline
\multirow{2}{*}{I4} & \multirow{2}{*}{Safari} & CloudTabs.db & 0 & 0 & 25 & 0 & 0 & \textbf{25} \\ \multirow{2}{*}{I4} & \multirow{2}{*}{Safari} & CloudTabs.db & 0 & 0 & 0 & 0 & 0 & \textbf{0} \\
\cline{3-9} \cline{3-9}
& & History.db & 0 & 0 & 4 & 0 & 0 & \textbf{4} \\ & & History.db & 0 & 0 & 0 & 0 & 0 & \textbf{0} \\
\hline \hline
\multirow{2}{*}{I5} & \multirow{2}{*}{Calendar} & Calendar.sqlitedb & 1 & 0 & 0 & 0 & 0 & \textbf{1} \\ \multirow{2}{*}{I5} & \multirow{2}{*}{Calendar} & Calendar.sqlitedb & 1 & 0 & 0 & 1 & 0 & \textbf{2} \\
\cline{3-9} \cline{3-9}
& & Extras.db & 0 & 0 & 0 & 0 & 0 & \textbf{0} \\ & & Extras.db & 0 & 0 & 0 & 0 & 0 & \textbf{0} \\
\hline \hline

32
RQs/RQ1/RQ1_t4_plain.csv
View File

@@ -1,26 +1,26 @@
ID,Application,Database,Email,Phone,UserName,PersonName,PostalAddr,Total ID,Application,Database,Email,Phone,UserName,PersonName,PostalAddr,Total
A1,WhatsApp,commerce.db,0,0,0,0,0,0 A1,WhatsApp,commerce.db,0,0,0,0,0,0
A1,WhatsApp,msgstore.db,0,7,5,15,0,27 A1,WhatsApp,msgstore.db,0,0,3,26,0,29
A1,WhatsApp,wa.db,0,16,680,10,0,706 A1,WhatsApp,wa.db,0,408,0,129,0,537
A2,Snapchat,core.db,0,1,5,1,0,7 A2,Snapchat,core.db,1,1,1,0,0,3
A2,Snapchat,journal.db,0,0,0,0,0,0 A2,Snapchat,journal.db,0,0,0,0,0,0
A2,Snapchat,main.db,1,13,0,11,0,25 A2,Snapchat,main.db,1,13,1229,252,0,1495
A3,Telegram,account1cache4.db,0,0,0,0,0,0 A3,Telegram,account1cache4.db,0,0,0,0,0,0
A3,Telegram,account2cache4.db,0,0,0,0,0,0 A3,Telegram,account2cache4.db,0,0,0,0,0,0
A3,Telegram,account3cache4.db,0,0,0,0,0,0 A3,Telegram,account3cache4.db,0,0,0,0,0,0
A4,Google Maps,gmm_myplaces.db,0,0,0,0,0,0 A4,Google Maps,gmm_myplaces.db,0,0,0,0,0,0
A4,Google Maps,gmm_storage.db,0,0,0,0,0,0 A4,Google Maps,gmm_storage.db,0,0,0,0,0,0
A4,Google Maps,peopleCache_sh....db,2,0,2,0,0,4 A4,Google Maps,peopleCache_sh....db,2,1,2,2,0,7
A5,Samsung Internet,SBrowser.db,1,0,5,0,0,6 A5,Samsung Internet,SBrowser.db,1,0,0,2,0,3
A5,Samsung Internet,SBrowser2.db,0,0,0,0,0,0 A5,Samsung Internet,SBrowser2.db,0,0,0,1,0,1
A5,Samsung Internet,searchengine.db,0,0,17,0,0,17 A5,Samsung Internet,searchengine.db,0,0,0,0,0,0
I1,WhatsApp,CallHistory.sqlite,0,0,10,0,0,10 I1,WhatsApp,CallHistory.sqlite,0,3,0,0,0,3
I1,WhatsApp,ChatStorage.sqlite,0,0,0,14,2,16 I1,WhatsApp,ChatStorage.sqlite,0,63,0,66,2,131
I1,WhatsApp,ContactsV2.sqlite,0,655,0,0,0,655 I1,WhatsApp,ContactsV2.sqlite,0,0,0,1350,0,1350
I2,Contacts,AddressBook.sqlitedb,6,548,1,748,0,1303 I2,Contacts,AddressBook.sqlitedb,6,716,0,733,0,1455
I2,Contacts,AddressB....sqlitedb,0,0,0,0,0,0 I2,Contacts,AddressB....sqlitedb,0,0,0,0,0,0
I3,Apple Messages,sms.db,1,0,0,10,0,11 I3,Apple Messages,sms.db,7,0,0,14,0,21
I4,Safari,CloudTabs.db,0,0,25,0,0,25 I4,Safari,CloudTabs.db,0,0,0,0,0,0
I4,Safari,History.db,0,0,4,0,0,4 I4,Safari,History.db,0,0,0,0,0,0
I5,Calendar,Calendar.sqlitedb,1,0,0,0,0,1 I5,Calendar,Calendar.sqlitedb,1,0,0,1,0,2
I5,Calendar,Extras.db,0,0,0,0,0,0 I5,Calendar,Extras.db,0,0,0,0,0,0
1 ID Application Database Email Phone UserName PersonName PostalAddr Total
2 A1 WhatsApp commerce.db 0 0 0 0 0 0
3 A1 WhatsApp msgstore.db 0 7 0 5 3 15 26 0 27 29
4 A1 WhatsApp wa.db 0 16 408 680 0 10 129 0 706 537
5 A2 Snapchat core.db 0 1 1 5 1 1 0 0 7 3
6 A2 Snapchat journal.db 0 0 0 0 0 0
7 A2 Snapchat main.db 1 13 0 1229 11 252 0 25 1495
8 A3 Telegram account1cache4.db 0 0 0 0 0 0
9 A3 Telegram account2cache4.db 0 0 0 0 0 0
10 A3 Telegram account3cache4.db 0 0 0 0 0 0
11 A4 Google Maps gmm_myplaces.db 0 0 0 0 0 0
12 A4 Google Maps gmm_storage.db 0 0 0 0 0 0
13 A4 Google Maps peopleCache_sh....db 2 0 1 2 0 2 0 4 7
14 A5 Samsung Internet SBrowser.db 1 0 5 0 0 2 0 6 3
15 A5 Samsung Internet SBrowser2.db 0 0 0 0 1 0 0 1
16 A5 Samsung Internet searchengine.db 0 0 17 0 0 0 17 0
17 I1 WhatsApp CallHistory.sqlite 0 0 3 10 0 0 0 10 3
18 I1 WhatsApp ChatStorage.sqlite 0 0 63 0 14 66 2 16 131
19 I1 WhatsApp ContactsV2.sqlite 0 655 0 0 0 1350 0 655 1350
20 I2 Contacts AddressBook.sqlitedb 6 548 716 1 0 748 733 0 1303 1455
21 I2 Contacts AddressB....sqlitedb 0 0 0 0 0 0
22 I3 Apple Messages sms.db 1 7 0 0 10 14 0 11 21
23 I4 Safari CloudTabs.db 0 0 25 0 0 0 25 0
24 I4 Safari History.db 0 0 4 0 0 0 4 0
25 I5 Calendar Calendar.sqlitedb 1 0 0 0 1 0 1 2
26 I5 Calendar Extras.db 0 0 0 0 0 0

12
RQs/RQ2/RQ2_search_space_reduction_gpt4o.tex
View File

@@ -7,23 +7,23 @@
\hline \hline
\textbf{ID} & \textbf{Apps} & \textbf{Candidate Cols (Total)} & \textbf{Cols Scanned (Extraction)} & \textbf{Reduc. (\%)} \\ \textbf{ID} & \textbf{Apps} & \textbf{Candidate Cols (Total)} & \textbf{Cols Scanned (Extraction)} & \textbf{Reduc. (\%)} \\
\hline \hline
A1 & WhatsApp & 1637 & 14 & 99.14\% \\ A1 & WhatsApp & 1637 & 7 & 99.57\% \\
\hline \hline
A2 & Snapchat & 848 & 2 & 99.76\% \\ A2 & Snapchat & 848 & 107 & 87.38\% \\
\hline \hline
A3 & Telegram & 1197 & 0 & 100.00\% \\ A3 & Telegram & 1197 & 0 & 100.00\% \\
\hline \hline
A4 & Google Maps & 80 & 2 & 97.50\% \\ A4 & Google Maps & 80 & 2 & 97.50\% \\
\hline \hline
A5 & Samsung Internet & 185 & 11 & 94.05\% \\ A5 & Samsung Internet & 185 & 0 & 100.00\% \\
\hline \hline
I1 & WhatsApp & 328 & 6 & 98.17\% \\ I1 & WhatsApp & 328 & 0 & 100.00\% \\
\hline \hline
I2 & Contacts & 13 & 13 & 0.00\% \\ I2 & Contacts & 13 & 0 & 100.00\% \\
\hline \hline
I3 & Apple Messages & 186 & 0 & 100.00\% \\ I3 & Apple Messages & 186 & 0 & 100.00\% \\
\hline \hline
I4 & Safari & 74 & 7 & 90.54\% \\ I4 & Safari & 74 & 0 & 100.00\% \\
\hline \hline
I5 & Calendar & 541 & 0 & 100.00\% \\ I5 & Calendar & 541 & 0 & 100.00\% \\
\hline \hline

16
RQs/RQ2/RQ2_t6_search_reduction.ipynb
View File

@@ -2,7 +2,7 @@
"cells": [ "cells": [
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 2, "execution_count": 3,
"id": "a30eef73", "id": "a30eef73",
"metadata": {}, "metadata": {},
"outputs": [ "outputs": [
@@ -12,15 +12,15 @@
"text": [ "text": [
"ID Apps CandidateCols ColsScanned Reduc(%)\n", "ID Apps CandidateCols ColsScanned Reduc(%)\n",
"-------------------------------------------------------------\n", "-------------------------------------------------------------\n",
"A1 WhatsApp 1637 14 99.14%\n", "A1 WhatsApp 1637 7 99.57%\n",
"A2 Snapchat 848 2 99.76%\n", "A2 Snapchat 848 107 87.38%\n",
"A3 Telegram 1197 0 100.00%\n", "A3 Telegram 1197 0 100.00%\n",
"A4 Google Maps 80 2 97.50%\n", "A4 Google Maps 80 2 97.50%\n",
"A5 Samsung Internet 185 11 94.05%\n", "A5 Samsung Internet 185 0 100.00%\n",
"I1 WhatsApp 328 6 98.17%\n", "I1 WhatsApp 328 0 100.00%\n",
"I2 Contacts 13 13 0.00%\n", "I2 Contacts 13 0 100.00%\n",
"I3 Apple Messages 186 0 100.00%\n", "I3 Apple Messages 186 0 100.00%\n",
"I4 Safari 74 7 90.54%\n", "I4 Safari 74 0 100.00%\n",
"I5 Calendar 541 0 100.00%\n", "I5 Calendar 541 0 100.00%\n",
"test2 test2 12 0 100.00%\n", "test2 test2 12 0 100.00%\n",
"users users 3 0 100.00%\n", "users users 3 0 100.00%\n",
@@ -38,7 +38,7 @@
"\n", "\n",
"CSV_PATH = Path(r\"app_total_columns.csv\")\n", "CSV_PATH = Path(r\"app_total_columns.csv\")\n",
"\n", "\n",
"BASE_DIR=Path(r\"..\\normalized_PII_results\\gpt4o\\app_level\")\n", "BASE_DIR=Path(r\"..\\normalized_PII_results\\GPT-5.1\\app_level\")\n",
"JSONL_PATH = BASE_DIR /Path(r\"app_level.jsonl\")\n", "JSONL_PATH = BASE_DIR /Path(r\"app_level.jsonl\")\n",
"OUT_TEX = Path(\"RQ2_search_space_reduction_gpt4o.tex\")\n", "OUT_TEX = Path(\"RQ2_search_space_reduction_gpt4o.tex\")\n",
"\n", "\n",

216
RQs/RQ3/RQ3_t10.ipynb
View File

@@ -9,7 +9,7 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": null, "execution_count": 1,
"metadata": {}, "metadata": {},
"outputs": [], "outputs": [],
"source": [ "source": [
@@ -23,7 +23,7 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": null, "execution_count": 2,
"metadata": {}, "metadata": {},
"outputs": [], "outputs": [],
"source": [ "source": [
@@ -37,7 +37,7 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": null, "execution_count": 3,
"metadata": {}, "metadata": {},
"outputs": [], "outputs": [],
"source": [ "source": [
@@ -97,7 +97,7 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": null, "execution_count": 4,
"metadata": {}, "metadata": {},
"outputs": [], "outputs": [],
"source": [ "source": [
@@ -122,7 +122,7 @@
" if not gt_set:\n", " if not gt_set:\n",
" row[col_name] = '-'\n", " row[col_name] = '-'\n",
" else:\n", " else:\n",
" # Ra,t = |Ga,t \u2229 Sa,t| / |Ga,t|\n", " # Ra,t = |Ga,t ∩ Sa,t| / |Ga,t|\n",
" recall = len(gt_set.intersection(system_set)) / len(gt_set)\n", " recall = len(gt_set.intersection(system_set)) / len(gt_set)\n",
" row[col_name] = recall\n", " row[col_name] = recall\n",
"\n", "\n",
@@ -130,7 +130,7 @@
" if not all_gt_pii_for_app:\n", " if not all_gt_pii_for_app:\n",
" row['All PII'] = '-'\n", " row['All PII'] = '-'\n",
" else:\n", " else:\n",
" # Ra,all = |(U Ga,t) \u2229 (U Sa,t)| / |U Ga,t|\n", " # Ra,all = |(U Ga,t) ∩ (U Sa,t)| / |U Ga,t|\n",
" all_recall = len(all_gt_pii_for_app.intersection(all_system_pii_for_app)) / len(all_gt_pii_for_app)\n", " all_recall = len(all_gt_pii_for_app.intersection(all_system_pii_for_app)) / len(all_gt_pii_for_app)\n",
" row['All PII'] = all_recall\n", " row['All PII'] = all_recall\n",
" \n", " \n",
@@ -139,9 +139,174 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": null, "execution_count": 5,
"metadata": {}, "metadata": {},
"outputs": [], "outputs": [
{
"data": {
"text/html": [
"<div>\n",
"<style scoped>\n",
" .dataframe tbody tr th:only-of-type {\n",
" vertical-align: middle;\n",
" }\n",
"\n",
" .dataframe tbody tr th {\n",
" vertical-align: top;\n",
" }\n",
"\n",
" .dataframe thead th {\n",
" text-align: right;\n",
" }\n",
"</style>\n",
"<table border=\"1\" class=\"dataframe\">\n",
" <thead>\n",
" <tr style=\"text-align: right;\">\n",
" <th></th>\n",
" <th>Application</th>\n",
" <th>Email</th>\n",
" <th>Phone</th>\n",
" <th>User Name</th>\n",
" <th>Person Name</th>\n",
" <th>Postal Address</th>\n",
" <th>All PII</th>\n",
" </tr>\n",
" <tr>\n",
" <th>ID</th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
" <tr>\n",
" <th>A1</th>\n",
" <td>WhatsApp</td>\n",
" <td>-</td>\n",
" <td>0.96</td>\n",
" <td>0.50</td>\n",
" <td>0.68</td>\n",
" <td>-</td>\n",
" <td>0.79</td>\n",
" </tr>\n",
" <tr>\n",
" <th>A2</th>\n",
" <td>Snapchat</td>\n",
" <td>1.00</td>\n",
" <td>1.00</td>\n",
" <td>0.33</td>\n",
" <td>1.00</td>\n",
" <td>-</td>\n",
" <td>0.79</td>\n",
" </tr>\n",
" <tr>\n",
" <th>A3</th>\n",
" <td>Telegram</td>\n",
" <td>-</td>\n",
" <td>-</td>\n",
" <td>-</td>\n",
" <td>-</td>\n",
" <td>-</td>\n",
" <td>-</td>\n",
" </tr>\n",
" <tr>\n",
" <th>A4</th>\n",
" <td>Google Maps</td>\n",
" <td>1.00</td>\n",
" <td>-</td>\n",
" <td>1.00</td>\n",
" <td>-</td>\n",
" <td>-</td>\n",
" <td>1.00</td>\n",
" </tr>\n",
" <tr>\n",
" <th>A5</th>\n",
" <td>Samsung Internet</td>\n",
" <td>1.00</td>\n",
" <td>-</td>\n",
" <td>0.00</td>\n",
" <td>-</td>\n",
" <td>-</td>\n",
" <td>0.20</td>\n",
" </tr>\n",
" <tr>\n",
" <th>I1</th>\n",
" <td>WhatsApp (iOS)</td>\n",
" <td>-</td>\n",
" <td>-</td>\n",
" <td>-</td>\n",
" <td>1.00</td>\n",
" <td>1.00</td>\n",
" <td>1.00</td>\n",
" </tr>\n",
" <tr>\n",
" <th>I2</th>\n",
" <td>Contacts</td>\n",
" <td>1.00</td>\n",
" <td>0.47</td>\n",
" <td>-</td>\n",
" <td>0.86</td>\n",
" <td>-</td>\n",
" <td>0.65</td>\n",
" </tr>\n",
" <tr>\n",
" <th>I3</th>\n",
" <td>Apple Messages</td>\n",
" <td>1.00</td>\n",
" <td>0.00</td>\n",
" <td>0.00</td>\n",
" <td>1.00</td>\n",
" <td>-</td>\n",
" <td>0.33</td>\n",
" </tr>\n",
" <tr>\n",
" <th>I4</th>\n",
" <td>Safari</td>\n",
" <td>-</td>\n",
" <td>-</td>\n",
" <td>0.02</td>\n",
" <td>-</td>\n",
" <td>-</td>\n",
" <td>0.02</td>\n",
" </tr>\n",
" <tr>\n",
" <th>I5</th>\n",
" <td>Calendar</td>\n",
" <td>1.00</td>\n",
" <td>-</td>\n",
" <td>-</td>\n",
" <td>-</td>\n",
" <td>-</td>\n",
" <td>1.00</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"</div>"
],
"text/plain": [
" Application Email Phone User Name Person Name Postal Address All PII\n",
"ID \n",
"A1 WhatsApp - 0.96 0.50 0.68 - 0.79\n",
"A2 Snapchat 1.00 1.00 0.33 1.00 - 0.79\n",
"A3 Telegram - - - - - -\n",
"A4 Google Maps 1.00 - 1.00 - - 1.00\n",
"A5 Samsung Internet 1.00 - 0.00 - - 0.20\n",
"I1 WhatsApp (iOS) - - - 1.00 1.00 1.00\n",
"I2 Contacts 1.00 0.47 - 0.86 - 0.65\n",
"I3 Apple Messages 1.00 0.00 0.00 1.00 - 0.33\n",
"I4 Safari - - 0.02 - - 0.02\n",
"I5 Calendar 1.00 - - - - 1.00"
]
},
"execution_count": 5,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [ "source": [
"df = pd.DataFrame(table_data)\n", "df = pd.DataFrame(table_data)\n",
"\n", "\n",
@@ -162,9 +327,38 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": null, "execution_count": 6,
"metadata": {}, "metadata": {},
"outputs": [], "outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\\begin{table}\n",
"\\caption{Per-application distinct recall.}\n",
"\\label{tab:app_level_recall}\n",
"\\begin{tabular}{llllllll}\n",
"\\toprule\n",
" & Application & Email & Phone & User Name & Person Name & Postal Address & All PII \\\\\n",
"ID & & & & & & & \\\\\n",
"\\midrule\n",
"A1 & WhatsApp & - & 0.96 & 0.50 & 0.68 & - & 0.79 \\\\\n",
"A2 & Snapchat & 1.00 & 1.00 & 0.33 & 1.00 & - & 0.79 \\\\\n",
"A3 & Telegram & - & - & - & - & - & - \\\\\n",
"A4 & Google Maps & 1.00 & - & 1.00 & - & - & 1.00 \\\\\n",
"A5 & Samsung Internet & 1.00 & - & 0.00 & - & - & 0.20 \\\\\n",
"I1 & WhatsApp (iOS) & - & - & - & 1.00 & 1.00 & 1.00 \\\\\n",
"I2 & Contacts & 1.00 & 0.47 & - & 0.86 & - & 0.65 \\\\\n",
"I3 & Apple Messages & 1.00 & 0.00 & 0.00 & 1.00 & - & 0.33 \\\\\n",
"I4 & Safari & - & - & 0.02 & - & - & 0.02 \\\\\n",
"I5 & Calendar & 1.00 & - & - & - & - & 1.00 \\\\\n",
"\\bottomrule\n",
"\\end{tabular}\n",
"\\end{table}\n",
"\n"
]
}
],
"source": [ "source": [
"# Optional: Save to LaTeX\n", "# Optional: Save to LaTeX\n",
"latex_output = df.to_latex(index=True, caption='Per-application distinct recall.', label='tab:app_level_recall', na_rep='-')\n", "latex_output = df.to_latex(index=True, caption='Per-application distinct recall.', label='tab:app_level_recall', na_rep='-')\n",
@@ -193,4 +387,4 @@
}, },
"nbformat": 4, "nbformat": 4,
"nbformat_minor": 4 "nbformat_minor": 4
} }

4
RQs/RQ3/RQ3_t7_corpus_level_Recall_Perc.ipynb
View File

@@ -2,7 +2,7 @@
"cells": [ "cells": [
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 3, "execution_count": null,
"id": "234eed3f", "id": "234eed3f",
"metadata": {}, "metadata": {},
"outputs": [ "outputs": [
@@ -35,7 +35,7 @@
"from pathlib import Path\n", "from pathlib import Path\n",
"from typing import Dict, Set, List\n", "from typing import Dict, Set, List\n",
"\n", "\n",
"SYSTEM_PATH = Path(r\"..\\normalized_PII_results\\gpt4o\\corpus_level\\corpus_level.jsonl\")\n", "SYSTEM_PATH = Path(r\"..\\normalized_PII_results\\GPT-5.1\\corpus_level\\corpus_level.jsonl\")\n",
"GT_PATH = Path(r\"..\\normalized_PII_results\\ground_truth\\corpus_level\\corpus_level.jsonl\")\n", "GT_PATH = Path(r\"..\\normalized_PII_results\\ground_truth\\corpus_level\\corpus_level.jsonl\")\n",
"\n", "\n",
"PII_TYPE_ORDER = [\"EMAIL\", \"PHONE\", \"USERNAME\", \"PERSON_NAME\", \"POSTAL_ADDRESS\"]\n", "PII_TYPE_ORDER = [\"EMAIL\", \"PHONE\", \"USERNAME\", \"PERSON_NAME\", \"POSTAL_ADDRESS\"]\n",

6
RQs/RQ3/RQ_t8.ipynb → RQs/RQ3/RQ3_t8.ipynb
View File

@@ -2,7 +2,7 @@
"cells": [ "cells": [
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 1, "execution_count": null,
"id": "1affac71", "id": "1affac71",
"metadata": {}, "metadata": {},
"outputs": [ "outputs": [
@@ -195,8 +195,8 @@
"\n", "\n",
"def main() -> None:\n", "def main() -> None:\n",
" # Define these inside main so importing this module has no side effects.\n", " # Define these inside main so importing this module has no side effects.\n",
" SYSTEM_DIR = Path(r\"..\\normalized_PII_results\\gpt4o\\db_level\")\n", " SYSTEM_DIR = Path(r\"..\\normalized_PII_results\\GPT-5.1\\db_level\")\n",
" GT_DIR = Path(r\"..\\normalized_PII_results\\gpt4o\\app_level\")\n", " GT_DIR = Path(r\"..\\normalized_PII_results\\GPT-5.1\\app_level\")\n",
" \n", " \n",
" gt_sets = collect_db_sets(GT_DIR, [k for k, _ in PII_TYPES])\n", " gt_sets = collect_db_sets(GT_DIR, [k for k, _ in PII_TYPES])\n",
" sys_sets = collect_db_sets(SYSTEM_DIR, [k for k, _ in PII_TYPES])\n", " sys_sets = collect_db_sets(SYSTEM_DIR, [k for k, _ in PII_TYPES])\n",

BIN
RQs/__pycache__/config.cpython-310.pyc Normal file
View File

Binary file not shown.

BIN
RQs/__pycache__/stats_utils.cpython-310.pyc
View File

Binary file not shown.

2
RQs/config.py
View File

@@ -1,4 +1,4 @@
GPT4O_RESULTS_DIR = 'normalized_PII_results/gpt4o/db_level/' GPT4O_RESULTS_DIR = 'normalized_PII_results/GPT-5.1/db_level/'
GROUND_TRUTH_DIR = 'normalized_PII_results/ground_truth/db_level/' GROUND_TRUTH_DIR = 'normalized_PII_results/ground_truth/db_level/'
PII_TYPES = ['EMAIL', 'PHONE', 'USERNAME', 'PERSON_NAME', 'POSTAL_ADDRESS'] PII_TYPES = ['EMAIL', 'PHONE', 'USERNAME', 'PERSON_NAME', 'POSTAL_ADDRESS']

50
RQs/normalized_PII_results/GPT-5.1/app_level/app_level.jsonl Normal file
View File

File diff suppressed because one or more lines are too long

5
RQs/normalized_PII_results/GPT-5.1/corpus_level/corpus_level.jsonl Normal file
View File

File diff suppressed because one or more lines are too long

0
RQs/normalized_PII_results/ground_truth/db_level/PII_A1_commerce_20260127T175911Z.jsonl → RQs/normalized_PII_results/GPT-5.1/db_level/PII_A1_commerce_20260211T022802Z.jsonl
View File

5
RQs/normalized_PII_results/GPT-5.1/db_level/PII_A1_msgstore_20260211T024003Z.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "USERNAME", "PII": ["status", "lid_me", "status_me"], "Num_of_PII": 3, "source_columns": ["jid.user", "jid.raw_string", "chat.subject", "message_text.description", "message_text.page_title", "message_text.url", "message_text.counter_abuse_token"], "Num_of_source_columns": 7}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "PERSON_NAME", "PII": ["mary garcia", "jim anderson", "professor jim", "mr. jim", "ms. mary", "mr. wilson", "abe", "matt", "gary gensler", "jerome powell", "benjamin melman", "cathy wood", "harvey jones", "michael saylor", "janet yellen", "jimmy butler", "lebron james", "michael jordan", "lionel messi", "biden", "mccarthy", "ron desantis", "jennifer farer", "amy berman jackson", "moynihan", "abbas al qattan"], "Num_of_PII": 26, "source_columns": ["jid.user", "jid.raw_string", "message.text_data", "message.translated_text", "chat.subject", "message_text.description", "message_text.page_title", "message_vcard.vcard"], "Num_of_source_columns": 8}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/GPT-5.1/db_level/PII_A1_wa_20260211T024706Z.jsonl Normal file
View File

File diff suppressed because one or more lines are too long

5
RQs/normalized_PII_results/GPT-5.1/db_level/PII_A2_core_20260211T023156Z.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "EMAIL", "PII": ["sharononeil368@gmail.com"], "Num_of_PII": 1, "source_columns": ["configetag.etag", "configrule.config_id", "deltaforcesync.client_key", "deltaforcesync.group_key", "preferences.stringvalue", "snapuserstore.groupkey", "snapuserstore.textval", "snapchatuserproperties.textval"], "Num_of_source_columns": 8}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "PHONE", "PII": ["8624338329"], "Num_of_PII": 1, "source_columns": ["configetag.etag", "configrule.config_id", "deltaforcesync.client_key", "preferences.key", "preferences.stringvalue", "snapuserstore.groupkey", "snapuserstore.textval", "snapchatuserproperties.textval"], "Num_of_source_columns": 8}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "USERNAME", "PII": ["oneil3607"], "Num_of_PII": 1, "source_columns": ["preferences.stringvalue", "snapuserstore.textval", "snapchatuserproperties.textval"], "Num_of_source_columns": 3}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "PERSON_NAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

0
RQs/normalized_PII_results/ground_truth/db_level/PII_A2_journal_20260127T180440Z.jsonl → RQs/normalized_PII_results/GPT-5.1/db_level/PII_A2_journal_20260211T023216Z.jsonl
View File

5
RQs/normalized_PII_results/GPT-5.1/db_level/PII_A2_main_20260211T023611Z.jsonl Normal file
View File

File diff suppressed because one or more lines are too long

0
RQs/normalized_PII_results/ground_truth/db_level/PII_A3_account1cache4_20260127T180745Z.jsonl → RQs/normalized_PII_results/GPT-5.1/db_level/PII_A3_account1cache4_20260211T023627Z.jsonl
View File

0
RQs/normalized_PII_results/ground_truth/db_level/PII_A3_account2cache4_20260127T180821Z.jsonl → RQs/normalized_PII_results/GPT-5.1/db_level/PII_A3_account2cache4_20260211T023643Z.jsonl
View File

0
RQs/normalized_PII_results/ground_truth/db_level/PII_A3_account3cache4_20260127T180857Z.jsonl → RQs/normalized_PII_results/GPT-5.1/db_level/PII_A3_account3cache4_20260211T023659Z.jsonl
View File

0
RQs/normalized_PII_results/ground_truth/db_level/PII_A4_gmm_myplaces_20260127T180935Z.jsonl → RQs/normalized_PII_results/GPT-5.1/db_level/PII_A4_gmm_myplaces_20260211T025539Z.jsonl
View File

0
RQs/normalized_PII_results/ground_truth/db_level/PII_A4_gmm_storage_20260127T181014Z.jsonl → RQs/normalized_PII_results/GPT-5.1/db_level/PII_A4_gmm_storage_20260211T025558Z.jsonl
View File

5
RQs/normalized_PII_results/GPT-5.1/db_level/PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14_20260211T025625Z.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "EMAIL", "PII": ["heather@cellebrite.com", "hmahalik@gmail.com"], "Num_of_PII": 2, "source_columns": ["tokens.value", "tokens_content.c1value"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "PHONE", "PII": ["7423794330"], "Num_of_PII": 1, "source_columns": ["tokens_content.c1value"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "USERNAME", "PII": ["heather", "hmahalik"], "Num_of_PII": 2, "source_columns": ["tokens.value", "tokens_content.c1value"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "PERSON_NAME", "PII": ["heather", "bo"], "Num_of_PII": 2, "source_columns": ["tokens.value", "tokens_content.c1value"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

2
RQs/normalized_PII_results/ground_truth/db_level/PII_A5_SBrowser2_20260127T181345Z.jsonl → RQs/normalized_PII_results/GPT-5.1/db_level/PII_A5_SBrowser2_20260211T025812Z.jsonl
View File

@@ -1,5 +1,5 @@
{"db_path": "selectedDBs\\A5_SBrowser2.db", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0} {"db_path": "selectedDBs\\A5_SBrowser2.db", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_SBrowser2.db", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0} {"db_path": "selectedDBs\\A5_SBrowser2.db", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_SBrowser2.db", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0} {"db_path": "selectedDBs\\A5_SBrowser2.db", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_SBrowser2.db", "PII_type": "PERSON_NAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0} {"db_path": "selectedDBs\\A5_SBrowser2.db", "PII_type": "PERSON_NAME", "PII": ["sharon"], "Num_of_PII": 1, "source_columns": ["bookmarks.title", "bookmarks.tags", "bookmarks.account_name", "bookmarks.device_name", "bookmarks.keyword", "bookmarks.description"], "Num_of_source_columns": 6}
{"db_path": "selectedDBs\\A5_SBrowser2.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0} {"db_path": "selectedDBs\\A5_SBrowser2.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/GPT-5.1/db_level/PII_A5_SBrowser_20260211T025741Z.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A5_SBrowser.db", "PII_type": "EMAIL", "PII": ["sharononeil368@gmail.com"], "Num_of_PII": 1, "source_columns": ["bookmarks.url", "bookmarks.surl", "bookmarks.title", "bookmarks.tags", "bookmarks.sourceid", "bookmarks.account_name", "bookmarks.account_type", "bookmarks.device_id", "bookmarks.device_name", "bookmarks.sync1", "bookmarks.sync2", "bookmarks.sync3", "bookmarks.sync4", "bookmarks.sync5", "bookmarks.keyword", "bookmarks.description", "bookmarks.guid", "sync_state.account_name", "sync_state.account_type", "sync_state.data", "tabs.tab_url", "tabs.tab_title", "tabs.account_name", "tabs.account_type", "tabs.sync1", "tabs.sync2", "tabs.sync3", "tabs.sync4", "tabs.sync5", "tabs.device_name", "tabs.device_id", "tabs.device_type", "tabs.client_guid", "tabs.history", "tabs.favicon", "tabs.tab_usage", "tabs.tab_group_name"], "Num_of_source_columns": 37}
{"db_path": "selectedDBs\\A5_SBrowser.db", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_SBrowser.db", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_SBrowser.db", "PII_type": "PERSON_NAME", "PII": ["sharon o'neil", "sharon"], "Num_of_PII": 2, "source_columns": ["bookmarks.title", "bookmarks.tags", "bookmarks.account_name", "bookmarks.device_name", "bookmarks.keyword", "bookmarks.description", "tabs.account_name", "tabs.tab_title", "tabs.device_name", "tabs.tab_group_name"], "Num_of_source_columns": 10}
{"db_path": "selectedDBs\\A5_SBrowser.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

0
RQs/normalized_PII_results/ground_truth/db_level/PII_A5_searchengine_20260127T181446Z.jsonl → RQs/normalized_PII_results/GPT-5.1/db_level/PII_A5_searchengine_20260211T025835Z.jsonl
View File

4
RQs/normalized_PII_results/ground_truth/db_level/PII_I1_CallHistory_20260127T181557Z.jsonl → RQs/normalized_PII_results/GPT-5.1/db_level/PII_I1_CallHistory_20260211T025226Z.jsonl
View File

@@ -1,5 +1,5 @@
{"db_path": "selectedDBs\\I1_CallHistory.sqlite", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": ["zwaaggregatecallevent.zlinktoken", "zwacdcallevent.zcallidstring", "zwacdcallevent.zgroupcallcreatoruserjidstring", "zwacdcallevent.zgroupjidstring", "zwacdcalleventparticipant.zjidstring"], "Num_of_source_columns": 5} {"db_path": "selectedDBs\\I1_CallHistory.sqlite", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_CallHistory.sqlite", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": ["zwacdcallevent.zcallidstring", "zwacdcalleventparticipant.zjidstring", "zwaaggregatecallevent.zlinktoken"], "Num_of_source_columns": 3} {"db_path": "selectedDBs\\I1_CallHistory.sqlite", "PII_type": "PHONE", "PII": ["9735203731", "4847353029", "9199037779"], "Num_of_PII": 3, "source_columns": ["zwaaggregatecallevent.zlinktoken", "zwacdcallevent.zcallidstring", "zwacdcallevent.zgroupcallcreatoruserjidstring", "zwacdcallevent.zgroupjidstring", "zwacdcallevent.zscheduledid", "zwacdcalleventparticipant.zjidstring"], "Num_of_source_columns": 6}
{"db_path": "selectedDBs\\I1_CallHistory.sqlite", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0} {"db_path": "selectedDBs\\I1_CallHistory.sqlite", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_CallHistory.sqlite", "PII_type": "PERSON_NAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0} {"db_path": "selectedDBs\\I1_CallHistory.sqlite", "PII_type": "PERSON_NAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_CallHistory.sqlite", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0} {"db_path": "selectedDBs\\I1_CallHistory.sqlite", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/GPT-5.1/db_level/PII_I1_ChatStorage_20260211T025417Z.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "PHONE", "PII": ["2484345508", "1971678670", "1201846697", "1202713209", "1203706983", "1214940093", "1224436196", "1262286547", "1304939557", "1315779050", "1329210838", "1334475192", "1341213345", "1346298524", "1346631169", "1386937021", "1406410636", "1423353884", "1424524593", "1425979629", "1440983872", "1443523052", "1443657333", "1475225173", "1484735302", "1507391856", "1507432918", "1509387979", "1513252325", "1516287992", "1540397794", "1562338897", "1570238327", "1571267978", "1585666327", "1608724264", "1614588736", "1626217400", "1661502844", "1662219135", "1716247706", "1757202662", "1762524347", "1762524348", "1762524355", "1763357105", "1781715367", "1785253308", "1814280888", "1816385272", "1817532867", "1829765983", "1831348174", "1835216100", "1845519029", "1850716754", "1856357126", "1862433832", "1864765701", "1919903777", "1931207050", "1936331007", "1971267495"], "Num_of_PII": 63, "source_columns": ["zwachatsession.zcontactidentifier", "zwachatsession.zcontactjid", "zwachatsession.zpartnername", "zwagroupmember.zcontactidentifier", "zwagroupmember.zcontactname", "zwagroupmember.zfirstname", "zwagroupmember.zmemberjid", "zwagroupmemberschange.zgroupjid", "zwagroupmemberschange.zmemberjids", "zwamediaitem.zauthorname", "zwamediaitem.zvcardname", "zwamediaitem.zvcardstring", "zwamessage.zfromjid", "zwamessage.zpushname", "zwamessage.ztext", "zwamessage.ztojid", "zwamessagedataitem.zchatjid", "zwamessagedataitem.zcontent1", "zwamessagedataitem.zcontent2", "zwamessagedataitem.zmatchedtext", "zwamessagedataitem.zsenderjid", "zwamessagedataitem.zsummary", "zwamessagedataitem.ztitle", "zwagroupinfo.zcreatorjid", "zwagroupinfo.zownerjid", "zwagroupinfo.zsourcejid", "zwagroupinfo.zsubjectownerjid", "zwaprofilepictureitem.zjid", "zwaprofilepushname.zpushname", "zwaprofilepushname.zjid", "zwachatpushconfig.zjid"], "Num_of_source_columns": 31}
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "PERSON_NAME", "PII": ["chad hunt", "toni yu", "charles finley", "ronen engler", "john raynolds", "jonathan reyes", "johnny good", "russell philby", "abe rudder", "rick roll", "andy sieg", "lisena gocaj", "monica", "ernesto torres cantú", "n. gregory mankiw", "david bailey", "donald trump", "robert f. kennedy jr.", "vivek ramaswamy", "cynthia lummis", "jerome powell", "stephanie", "joe biden", "kamala harris", "matt gaetz", "mark cuban", "michael saylor", "christian", "sharon", "otto", "reynolds", "andy schweichert", "abner", "brian arseneau", "virginia benton", "howell", "yaliweisi", "jason", "sharon oneil", "job vizcarra", "robert elliott", "sultan", "emerick", "nia yuniar", "david wilson", "robechucks raul", "ella bella", "dick oscar", "charlie", "steven", "ameya joshi", "amiel williamson", "ashwin menon", "ajax edmiston", "eleazar lewden", "polly lucas", "lemuel glasgow", "bazel mcconnel", "william stevenson", "denice r allen", "leif fox", "william hopkins", "jonas bradley", "amit sharma", "finn", "colin dacopps"], "Num_of_PII": 66, "source_columns": ["zwachatsession.zpartnername", "zwachatsession.zcontactidentifier", "zwachatsession.zlastmessagetext", "zwachatsession.zsavedinput", "zwagroupmember.zcontactidentifier", "zwagroupmember.zcontactname", "zwagroupmember.zfirstname", "zwamediaitem.zauthorname", "zwamediaitem.zvcardname", "zwamessage.zpushname", "zwamessage.ztext", "zwamessagedataitem.zcontent1", "zwamessagedataitem.zcontent2", "zwamessagedataitem.zmatchedtext", "zwamessagedataitem.zsummary", "zwamessagedataitem.ztitle", "zwaprofilepushname.zpushname"], "Num_of_source_columns": 17}
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "POSTAL_ADDRESS", "PII": ["12503 e via de palmas, chandler, az", "8500 peña blvd, denver, co"], "Num_of_PII": 2, "source_columns": ["zwamessage.ztext", "zwamessagedataitem.zcontent1", "zwamessagedataitem.zcontent2", "zwamessagedataitem.zmatchedtext", "zwamessagedataitem.zsummary", "zwamessagedataitem.ztitle", "zwachatsession.zlastmessagetext"], "Num_of_source_columns": 7}

5
RQs/normalized_PII_results/GPT-5.1/db_level/PII_I1_ContactsV2_20260211T030947Z.jsonl Normal file
View File

File diff suppressed because one or more lines are too long

0
RQs/normalized_PII_results/ground_truth/db_level/PII_I2_AddressBookImages_20260127T183526Z.jsonl → RQs/normalized_PII_results/GPT-5.1/db_level/PII_I2_AddressBookImages_20260211T030157Z.jsonl
View File

5
RQs/normalized_PII_results/GPT-5.1/db_level/PII_I2_AddressBook_20260211T030143Z.jsonl Normal file
View File

File diff suppressed because one or more lines are too long

5
RQs/normalized_PII_results/GPT-5.1/db_level/PII_I3_sms_20260211T030251Z.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "EMAIL", "PII": ["aniulka616@hotmail.com", "josephphw3r5@outlook.com", "jreynoldscbdemo+this.is.secret@gmail.com", "jthskp131699@icloud.com", "sousiezorjrc@outlook.com", "sweeneyfrank594@gmail.com", "ottomatik1234@gmail.com"], "Num_of_PII": 7, "source_columns": ["handle.id", "handle.uncanonicalized_id", "chat.account_id", "chat.chat_identifier", "chat.account_login", "chat.last_addressed_handle", "chat.display_name", "chat.cloudkit_record_id", "attachment.filename", "attachment.transfer_name", "attachment.guid", "attachment.original_guid", "chat.guid", "chat.room_name", "chat.group_id", "chat.engram_id", "chat.server_change_token", "chat.original_group_id", "message.text", "message.service_center", "message.subject", "message.service", "message.account", "message.account_guid", "message.guid", "message.associated_message_guid", "message.balloon_bundle_id", "message.expressive_send_style_id", "message.ck_record_id", "message.ck_record_change_tag", "message.destination_caller_id", "message.reply_to_guid", "message.thread_originator_guid", "message.thread_originator_part", "message.syndication_ranges", "message.synced_syndication_ranges", "message.bia_reference_id", "message.fallback_hash"], "Num_of_source_columns": 38}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "PERSON_NAME", "PII": ["anya", "william", "eddie", "otto", "tracy", "dimitra", "anna", "joe church", "sharon oneil", "ella rutman doligo", "ed venture", "ron serber", "ronen engler", "abe"], "Num_of_PII": 14, "source_columns": ["chat.display_name", "chat.chat_identifier", "chat.account_login", "chat.last_addressed_handle", "handle.id", "handle.uncanonicalized_id", "message.text", "message.subject", "message.cache_roomnames"], "Num_of_source_columns": 9}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

0
RQs/normalized_PII_results/ground_truth/db_level/PII_I4_CloudTabs_20260127T183643Z.jsonl → RQs/normalized_PII_results/GPT-5.1/db_level/PII_I4_CloudTabs_20260211T030321Z.jsonl
View File

5
RQs/normalized_PII_results/GPT-5.1/db_level/PII_I4_History_20260211T030358Z.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\I4_History.db", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I4_History.db", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I4_History.db", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I4_History.db", "PII_type": "PERSON_NAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I4_History.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/GPT-5.1/db_level/PII_I5_Calendar_20260211T030514Z.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "EMAIL", "PII": ["ottomatik1234@gmail.com"], "Num_of_PII": 1, "source_columns": ["alarm.email_address", "calendar.shared_owner_address", "calendar.self_identity_email", "calendar.owner_identity_email", "identity.address"], "Num_of_source_columns": 5}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "PERSON_NAME", "PII": ["otto matik"], "Num_of_PII": 1, "source_columns": ["calendar.title", "calendar.shared_owner_name", "calendar.shared_owner_address", "calendar.self_identity_id", "calendar.self_identity_email", "calendar.self_identity_phone_number", "calendar.owner_identity_id", "calendar.owner_identity_email", "calendar.owner_identity_phone_number", "calendar.notes", "calendar.last_sync_title", "calendaritem.summary", "calendaritem.description", "calendaritem.response_comment", "calendaritem.last_synced_response_comment", "calendaritem.url", "calendaritem.conference_url", "calendaritem.contact_identifier", "calendaritem.contact_name", "calendaritem.creator_identity", "calendaritem.creator_team_identity", "calendaritem.special_day", "category.name", "clientcursor.client_identifier", "clientsequence.client_identifier", "identity.display_name", "identity.address", "identity.first_name", "identity.last_name", "store.name", "store.notes", "store.owner_name", "_sqlitedatabaseproperties.key", "_sqlitedatabaseproperties.value"], "Num_of_source_columns": 34}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

0
RQs/normalized_PII_results/ground_truth/db_level/PII_I5_Extras_20260127T183857Z.jsonl → RQs/normalized_PII_results/GPT-5.1/db_level/PII_I5_Extras_20260211T030537Z.jsonl
View File

8
RQs/normalized_PII_results/gpt4o/app_level/app_level.jsonl
View File

File diff suppressed because one or more lines are too long

2
RQs/normalized_PII_results/gpt4o/corpus_level/corpus_level.jsonl
View File

File diff suppressed because one or more lines are too long

2
RQs/normalized_PII_results/gpt4o/db_level/PII_A1_msgstore_20260131T203502Z.jsonl
View File

@@ -1,5 +1,5 @@
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0} {"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "PHONE", "PII": ["2023133725", "9106995488", "14244990541", "14359905938", "16467602090", "13346095713", "17622338037"], "Num_of_PII": 7, "source_columns": ["message.text_data", "chat.subject", "call_log.call_id", "message_text.description"], "Num_of_source_columns": 4} {"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "PHONE", "PII": ["2023133725", "9106995488", "4244990541", "4359905938", "6467602090", "3346095713", "7622338037"], "Num_of_PII": 7, "source_columns": ["message.text_data", "chat.subject", "call_log.call_id", "message_text.description"], "Num_of_source_columns": 4}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "USERNAME", "PII": ["wealthbuildersclub", "bitcoinmastersjiminvestmentteam", "marygarcia", "jimanderson", "btcointradingcenter"], "Num_of_PII": 5, "source_columns": ["chat.subject", "message.text_data", "message_text.description", "user_device.user_jid_row_id", "user_device_info.user_jid_row_id"], "Num_of_source_columns": 5} {"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "USERNAME", "PII": ["wealthbuildersclub", "bitcoinmastersjiminvestmentteam", "marygarcia", "jimanderson", "btcointradingcenter"], "Num_of_PII": 5, "source_columns": ["chat.subject", "message.text_data", "message_text.description", "user_device.user_jid_row_id", "user_device_info.user_jid_row_id"], "Num_of_source_columns": 5}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "PERSON_NAME", "PII": ["jim investment team", "professor jim", "jim anderson", "mary garcia", "benjamin", "lorie logan", "biden", "yellen", "ron desantis", "michael saylor", "cathy wood", "harvey jones", "gary gensler", "abbas al qattan", "jw verret"], "Num_of_PII": 15, "source_columns": ["chat.subject", "message.text_data", "message_text.description", "message_vcard.vcard"], "Num_of_source_columns": 4} {"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "PERSON_NAME", "PII": ["jim investment team", "professor jim", "jim anderson", "mary garcia", "benjamin", "lorie logan", "biden", "yellen", "ron desantis", "michael saylor", "cathy wood", "harvey jones", "gary gensler", "abbas al qattan", "jw verret"], "Num_of_PII": 15, "source_columns": ["chat.subject", "message.text_data", "message_text.description", "message_vcard.vcard"], "Num_of_source_columns": 4}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0} {"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

2
RQs/normalized_PII_results/gpt4o/db_level/PII_A1_wa_20260131T203943Z.jsonl
View File

File diff suppressed because one or more lines are too long

2
RQs/normalized_PII_results/gpt4o/db_level/PII_A2_core_20260131T204055Z.jsonl
View File

@@ -1,5 +1,5 @@
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0} {"db_path": "selectedDBs\\A2_core.db", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "PHONE", "PII": ["18624338329"], "Num_of_PII": 1, "source_columns": ["configetag.etag", "configrule.config_id", "deltaforcesync.client_key", "preferences.key", "snapuserstore.textval", "snapchatuserproperties.textval", "android_metadata.locale"], "Num_of_source_columns": 7} {"db_path": "selectedDBs\\A2_core.db", "PII_type": "PHONE", "PII": ["8624338329"], "Num_of_PII": 1, "source_columns": ["configetag.etag", "configrule.config_id", "deltaforcesync.client_key", "preferences.key", "snapuserstore.textval", "snapchatuserproperties.textval", "android_metadata.locale"], "Num_of_source_columns": 7}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "USERNAME", "PII": ["oneil3607", "sharononeil368", "no_skin_tone", "static_image", "memories_and_camera_roll"], "Num_of_PII": 5, "source_columns": ["snapuserstore.textval", "snapchatuserproperties.textval"], "Num_of_source_columns": 2} {"db_path": "selectedDBs\\A2_core.db", "PII_type": "USERNAME", "PII": ["oneil3607", "sharononeil368", "no_skin_tone", "static_image", "memories_and_camera_roll"], "Num_of_PII": 5, "source_columns": ["snapuserstore.textval", "snapchatuserproperties.textval"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "PERSON_NAME", "PII": ["sharon oneil"], "Num_of_PII": 1, "source_columns": ["preferences.stringvalue", "snapuserstore.textval", "snapchatuserproperties.textval"], "Num_of_source_columns": 3} {"db_path": "selectedDBs\\A2_core.db", "PII_type": "PERSON_NAME", "PII": ["sharon oneil"], "Num_of_PII": 1, "source_columns": ["preferences.stringvalue", "snapuserstore.textval", "snapchatuserproperties.textval"], "Num_of_source_columns": 3}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0} {"db_path": "selectedDBs\\A2_core.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

2
RQs/normalized_PII_results/gpt4o/db_level/PII_I1_ContactsV2_20260131T210034Z.jsonl
View File

File diff suppressed because one or more lines are too long

2
RQs/normalized_PII_results/gpt4o/db_level/PII_I2_AddressBook_20260131T210607Z.jsonl
View File

File diff suppressed because one or more lines are too long

50
RQs/normalized_PII_results/ground_truth/app_level/app_level.jsonl
View File

File diff suppressed because one or more lines are too long

5
RQs/normalized_PII_results/ground_truth/corpus_level/corpus_level.jsonl
View File

File diff suppressed because one or more lines are too long

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A1_commerce.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A1_commerce.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A1_commerce.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A1_commerce.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A1_commerce.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A1_commerce.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A1_msgstore.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["available_message_view.text_data", "message.text_data", "message_ftsv2.content", "message_ftsv2_content.c0content", "message_system_business_state.business_name", "message_vcard.vcard"], "Num_of_source_columns": 6}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["jid.raw_string", "jid.user", "message.text_data"], "Num_of_source_columns": 3}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["available_message_view.text_data", "message.text_data", "message_ftsv2.content", "message_ftsv2_content.c0content", "message_quoted.text_data", "message_vcard.vcard"], "Num_of_source_columns": 6}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["available_message_view.text_data", "message.text_data", "message_ftsv2.content", "message_ftsv2_content.c0content", "message_quoted.text_data"], "Num_of_source_columns": 5}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A1_msgstore_20260127T180043Z.jsonl
View File

@@ -1,5 +0,0 @@
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "PHONE", "PII": ["2023133725", "9106995488", "14244990541", "14359905938", "16467602090", "13346095713", "17622338037"], "Num_of_PII": 7, "source_columns": ["message.text_data", "chat.subject", "call_log.call_id", "message_text.description"], "Num_of_source_columns": 4}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "USERNAME", "PII": ["wealthbuildersclub", "btcoinmastersjiminvestmentteam"], "Num_of_PII": 2, "source_columns": ["chat.subject", "message.text_data", "jid.user", "user_device.user_jid_row_id"], "Num_of_source_columns": 4}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "PERSON_NAME", "PII": ["jim investment team", "mary garcia", "jim anderson", "professor jim", "mr. jim", "lorie logan", "benjamin", "mary", "ron desantis", "cathy wood", "michael saylor", "gary gensler", "harvey jones", "abbas al qattan", "jennifer farer", "damodaran", "moe", "abe", "matt", "ruth", "wilson", "rebeca"], "Num_of_PII": 22, "source_columns": ["chat.subject", "message.text_data", "message_text.description", "message_vcard.vcard"], "Num_of_source_columns": 4}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": ["message.text_data", "message_text.description", "message_location.place_address", "message_vcard.vcard"], "Num_of_source_columns": 4}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A1_wa.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["wa_address_book.number", "wa_contacts.number"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["wa_bot_profiles.creator_name", "wa_contacts.wa_name", "wa_vnames.verified_name"], "Num_of_source_columns": 3}
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["wa_address_book.display_name", "wa_address_book.given_name+family_name", "wa_address_book.sort_name", "wa_contacts.given_name+family_name", "wa_contacts.sort_name"], "Num_of_source_columns": 5}
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A1_wa_20260127T180213Z.jsonl
View File

@@ -1,5 +0,0 @@
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "PHONE", "PII": ["5713298742", "5713349815", "8085096467", "8624338328", "7034241981", "2028177932", "5715917168", "2025692832", "6106046786", "6263678865", "2065937224", "9199037779", "8056377243", "9735203731", "8136743027", "7423794330"], "Num_of_PII": 16, "source_columns": ["wa_contacts.number", "wa_address_book.number"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "PERSON_NAME", "PII": ["svetlana chernoff", "ltc whalen", "capt don wayne", "karen tate", "brian reynolds", "goldie kahn", "vladamir stravinsky", "mary garcia", "abe rudder", "russ philby"], "Num_of_PII": 10, "source_columns": ["wa_contacts.display_name", "wa_contacts.given_name", "wa_contacts.family_name", "wa_contacts.nickname", "wa_address_book.display_name", "wa_address_book.given_name", "wa_address_book.family_name", "wa_address_book.nickname", "wa_biz_profiles.business_description", "wa_biz_profiles.location_name"], "Num_of_source_columns": 10}
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A2_core.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["snapuserstore.textval"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["snapuserstore.textval"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["snapuserstore.textval"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["snapuserstore.textval"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A2_core_20260127T180339Z.jsonl
View File

@@ -1,5 +0,0 @@
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "PHONE", "PII": ["18624338329"], "Num_of_PII": 1, "source_columns": ["configetag.etag", "configrule.config_id", "deltaforcesync.client_key", "preferences.key", "snapuserstore.textval", "snapchatuserproperties.textval"], "Num_of_source_columns": 6}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "USERNAME", "PII": ["oneil3607", "no_skin_tone", "static_image", "chqkedmxnde2nmq3zddlytu3ngyqag", "memories_and_camera_roll", "caegqokatayoqokatazaaggp", "cncbaridx8kb", "caego7zptqyoo7zptqy", "caiyascjvm+1bije0tm1bkdg0tm1bkjg0tm1blab", "camgqokatayo09lztqy", "caigqrzptqyo09lztqy", "wgiqag"], "Num_of_PII": 12, "source_columns": ["snapuserstore.textval", "snapchatuserproperties.textval"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "PERSON_NAME", "PII": ["sharon oneil"], "Num_of_PII": 1, "source_columns": ["configetag.etag", "preferences.stringvalue", "snapuserstore.textval", "snapchatuserproperties.textval"], "Num_of_source_columns": 4}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A2_journal.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A2_journal.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A2_journal.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A2_journal.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A2_journal.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A2_journal.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A2_main.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["contact.phone", "contact.rawphone"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["combinedusername.mutableusername", "combinedusername.originalusername", "discoverfeedfriendstoriesviewv2.frienddisplayname", "friend.displayname", "friend.serverdisplayname", "friend.username", "friendwithusername.displayname", "friendwithusername.serverdisplayname", "friendwithusername.usernameforsorting", "friendsfeedviewv2.displayinteractionuserusername", "friendsfeedviewv2.frienddisplayusername", "storysnap.captiontextdisplay"], "Num_of_source_columns": 12}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["contact.displayname", "friendsfeedviewv2.displayinteractionuserdisplayname", "story.displayname"], "Num_of_source_columns": 3}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A2_main_20260127T180710Z.jsonl
View File

@@ -1,5 +0,0 @@
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "EMAIL", "PII": ["copyright@snap.com"], "Num_of_PII": 1, "source_columns": ["billboardstrings.message", "combinedusername.originalusername", "contact.displayname", "notificationdata.userid", "story.displayname", "suggestedfriend.userid"], "Num_of_source_columns": 6}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "PHONE", "PII": ["2065937224", "8624338328", "9199037779", "8085096467", "5713298742", "2028177932", "2025692832", "5713349815", "5715917168", "6106046786", "6263678865", "8056377243", "7423794330"], "Num_of_PII": 13, "source_columns": ["contact.phone", "billboardstrings.stringkey", "legalagreementstrings.message", "story.displayname", "storysnap.displayname"], "Num_of_source_columns": 5}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "PERSON_NAME", "PII": ["abe rudder", "karen tate", "russ philby", "capt don wayne", "svetlana chernoff", "brian reynolds", "vladamir stravinsky", "ltc whalen", "goldie kahn", "joey", "mary garcia", "sharon oneil"], "Num_of_PII": 12, "source_columns": ["contact.displayname", "combinedusername.originalusername", "feed.specifiedname", "story.displayname", "storysnap.displayname", "suggestedfriend.suggestionreason"], "Num_of_source_columns": 6}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A3_account1cache4.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A3_account1cache4.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A3_account1cache4.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A3_account1cache4.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A3_account1cache4.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A3_account1cache4.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A3_account2cache4.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A3_account2cache4.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A3_account2cache4.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A3_account2cache4.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A3_account2cache4.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A3_account2cache4.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A3_account3cache4.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A3_account3cache4.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A3_account3cache4.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A3_account3cache4.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A3_account3cache4.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A3_account3cache4.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A4_gmm_myplaces.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A4_gmm_myplaces.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A4_gmm_myplaces.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A4_gmm_myplaces.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A4_gmm_myplaces.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A4_gmm_myplaces.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A4_gmm_storage.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A4_gmm_storage.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A4_gmm_storage.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A4_gmm_storage.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A4_gmm_storage.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A4_gmm_storage.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["tokens.value", "tokens_content.c1value"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["tokens.value", "tokens_content.c1value"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["tokens.value", "tokens_content.c1value"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["tokens.value", "tokens_content.c1value"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14_20260127T181121Z.jsonl
View File

@@ -1,5 +0,0 @@
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "EMAIL", "PII": ["heather@cellebrite.com", "hmahalik@gmail.com"], "Num_of_PII": 2, "source_columns": ["tokens.value", "tokens_content.c1value", "tokens_stat.value", "cacheinfo.affinity_response_context"], "Num_of_source_columns": 4}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": ["tokens.value", "tokens_content.c1value", "tokens_stat.value"], "Num_of_source_columns": 3}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "USERNAME", "PII": ["heather", "hmahalik"], "Num_of_PII": 2, "source_columns": ["cacheinfo.affinity_response_context", "contacts.type", "tokens.value", "tokens_content.c1value", "tokens_stat.value"], "Num_of_source_columns": 5}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "PERSON_NAME", "PII": [], "Num_of_PII": 0, "source_columns": ["tokens.value", "tokens_stat.value", "contacts.id"], "Num_of_source_columns": 3}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A5_SBrowser.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A5_SBrowser.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["bookmarks.account_name", "sync_state.account_name", "tabs.account_name"], "Num_of_source_columns": 3}
{"db_path": "selectedDBs\\A5_SBrowser.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_SBrowser.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_SBrowser.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_SBrowser.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A5_SBrowser2.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A5_SBrowser2.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_SBrowser2.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_SBrowser2.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_SBrowser2.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_SBrowser2.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A5_SBrowser_20260127T181239Z.jsonl
View File

@@ -1,5 +0,0 @@
{"db_path": "selectedDBs\\A5_SBrowser.db", "PII_type": "EMAIL", "PII": ["sharononeil368@gmail.com"], "Num_of_PII": 1, "source_columns": ["bookmarks.account_name", "internet_sync.sync_key", "sync_state.account_name", "tabs.tab_title", "android_metadata.locale"], "Num_of_source_columns": 5}
{"db_path": "selectedDBs\\A5_SBrowser.db", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_SBrowser.db", "PII_type": "USERNAME", "PII": ["syncinternetdata", "syncbookmarks", "syncopenpages", "syncsavedpages"], "Num_of_PII": 4, "source_columns": ["bookmarks.account_name", "tabs.account_name", "sync_state.account_name", "internet_sync.sync_key"], "Num_of_source_columns": 4}
{"db_path": "selectedDBs\\A5_SBrowser.db", "PII_type": "PERSON_NAME", "PII": [], "Num_of_PII": 0, "source_columns": ["bookmarks.title", "bookmarks.account_name", "bookmarks.account_type", "sync_state.data", "tabs.tab_title", "tabs.account_name", "tabs.account_type"], "Num_of_source_columns": 7}
{"db_path": "selectedDBs\\A5_SBrowser.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_A5_searchengine.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\A5_searchengine.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_searchengine.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_searchengine.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_searchengine.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\A5_searchengine.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_I1_CallHistory.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\I1_CallHistory.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_CallHistory.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_CallHistory.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_CallHistory.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_CallHistory.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_I1_ChatStorage.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zwachatsession.zpartnername", "zwaprofilepushname.zpushname"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zwamediaitem.ztitle", "zwamessage.ztext", "zwamessagedataitem.zsummary", "zwamessagedataitem.ztitle"], "Num_of_source_columns": 4}
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zwachatsession.zpartnername", "zwamediaitem.ztitle", "zwamessage.ztext", "zwamessagedataitem.zsummary", "zwamessagedataitem.ztitle", "zwaprofilepushname.zpushname"], "Num_of_source_columns": 6}
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zwamediaitem.ztitle", "zwamessage.ztext", "zwamessagedataitem.ztitle"], "Num_of_source_columns": 3}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_I1_ChatStorage_20260127T181731Z.jsonl
View File

@@ -1,5 +0,0 @@
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "PERSON_NAME", "PII": ["rick", "otto", "reynolds", "emily", "sharon", "lisena gocaj", "andy sieg", "christian justiniano", "david wilson", "robechucks raul", "abner", "nia yuniar", "william stevenson", "amit sharma"], "Num_of_PII": 14, "source_columns": ["zwamessage.ztext", "zwamessagedataitem.zcontent1", "zwaprofilepushname.zpushname"], "Num_of_source_columns": 3}
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "POSTAL_ADDRESS", "PII": ["12503 e via de palmas, chandler, az", "8500 peña blvd, denver, co"], "Num_of_PII": 2, "source_columns": ["zwamessage.ztext", "zwamessagedataitem.zcontent1", "zwamessagedataitem.zcontent2"], "Num_of_source_columns": 3}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_I1_ContactsV2.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\I1_ContactsV2.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_ContactsV2.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zwaaddressbookcontact.zphonenumber"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I1_ContactsV2.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_ContactsV2.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zwaaddressbookcontact.zfullname"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I1_ContactsV2.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_I1_ContactsV2_20260127T182906Z.jsonl
View File

@@ -1,5 +0,0 @@
{"db_path": "selectedDBs\\I1_ContactsV2.sqlite", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_ContactsV2.sqlite", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": ["zwaaddressbookcontact.zphonenumber", "zwaaddressbookcontact.zlocalizedphonenumber"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\I1_ContactsV2.sqlite", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I1_ContactsV2.sqlite", "PII_type": "PERSON_NAME", "PII": [], "Num_of_PII": 0, "source_columns": ["zwaaddressbookcontact.zfullname", "zwaaddressbookcontact.zgivenname", "zwaaddressbookcontact.zlastname", "zwaaddressbookcontact.znotes"], "Num_of_source_columns": 4}
{"db_path": "selectedDBs\\I1_ContactsV2.sqlite", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_I2_AddressBook.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\I2_AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["abmultivalue.value", "abpersonfulltextsearch_content.c17email"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\I2_AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["abmultivalue.value"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I2_AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I2_AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["abperson.first+last", "abpersonfulltextsearch_content.c0first+c1last+c2middle"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\I2_AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["abmultivalueentry.value", "abpersonfulltextsearch_content.c18address"], "Num_of_source_columns": 2}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_I2_AddressBookImages.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\I2_AddressBookImages.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I2_AddressBookImages.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I2_AddressBookImages.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I2_AddressBookImages.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I2_AddressBookImages.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_I2_AddressBook_20260127T183457Z.jsonl
View File

File diff suppressed because one or more lines are too long

5
RQs/normalized_PII_results/ground_truth/db_level/PII_I3_sms.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["chat.account_login", "chat.chat_identifier", "chat.guid", "handle.id", "message.account"], "Num_of_source_columns": 5}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["chat.account_login", "chat.chat_identifier", "chat.guid", "chat.last_addressed_handle", "handle.id", "handle.uncanonicalized_id", "message.account", "message.destination_caller_id"], "Num_of_source_columns": 8}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_I3_sms_20260127T183606Z.jsonl
View File

@@ -1,5 +0,0 @@
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "EMAIL", "PII": ["ottomatik1234@gmail.com"], "Num_of_PII": 1, "source_columns": ["message.text", "chat.account_login", "kvtable.key", "attachment.guid", "chat.display_name"], "Num_of_source_columns": 5}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "PHONE", "PII": ["12065937224", "12185715037", "12484345508", "13017157263", "14155346421", "14197574625", "14259796297", "14325353346", "14847353029", "15162879924", "17709195681", "17852533080", "17868720110", "18553965089", "18589330241", "18624338324", "18624338329", "19195796456", "19195796465", "19199037779"], "Num_of_PII": 20, "source_columns": ["message.text", "chat.chat_identifier"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "USERNAME", "PII": ["chat70863266579689223", "chat818965589567390604"], "Num_of_PII": 2, "source_columns": ["chat.chat_identifier", "handle.id", "message.text"], "Num_of_source_columns": 3}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "PERSON_NAME", "PII": ["anya", "william", "eddie v", "otto", "tracy", "anna", "ronen engler", "joe church", "ella rutman doligo", "sharon oneil"], "Num_of_PII": 10, "source_columns": ["chat.display_name", "handle.id", "message.text"], "Num_of_source_columns": 3}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_I4_CloudTabs.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\I4_CloudTabs.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I4_CloudTabs.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I4_CloudTabs.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I4_CloudTabs.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I4_CloudTabs.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_I4_History.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\I4_History.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I4_History.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I4_History.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I4_History.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I4_History.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_I4_History_20260127T183727Z.jsonl
View File

@@ -1,5 +0,0 @@
{"db_path": "selectedDBs\\I4_History.db", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I4_History.db", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I4_History.db", "PII_type": "USERNAME", "PII": ["q1146771", "q1155133", "q104851", "q46383", "q1066823", "q81809", "q190120", "q336532", "q108143", "q27686", "q1190598", "q185165", "q170448", "q13233", "q524679", "q1076874", "q2766", "q388483", "q3180957", "q570871", "q682144", "q861508", "q15709638", "q35855", "q1229287", "q1605306", "q281", "q1390866", "q19363883", "q259011", "q154168", "q977090", "q1060705", "q22866", "q1273800", "q603481", "q1809515", "q181479", "q273027", "q2047030", "q908923", "q47616", "q372852", "q705450", "q1988120", "q16939396", "q908666", "q241987", "q62458", "q747493", "q3107826", "17ebu7rghueusrylzr6u3ccjwmmqprk28k", "cached_sync_circle_size", "current_generation", "fetch_throttler_data", "last_maintenance_date", "last_synced_generation", "profile_server_map", "push_notifications_initialized", "push_throttler_data", "server_change_token", "sync_circle_size_retrieval_throttler_data", "sync_with_manatee_container"], "Num_of_PII": 63, "source_columns": ["history_event_listeners.listener_name", "history_items.id", "history_items_to_tags.history_item", "history_tags.identifier", "history_visits.title", "metadata.key"], "Num_of_source_columns": 6}
{"db_path": "selectedDBs\\I4_History.db", "PII_type": "PERSON_NAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I4_History.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_I5_Calendar.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["calendar.last_sync_title", "calendar.notes", "calendar.owner_identity_email", "calendar.self_identity_email", "calendar.shared_owner_address", "calendar.title", "identity.address"], "Num_of_source_columns": 7}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["calendar.shared_owner_name", "identity.display_name", "store.owner_name"], "Num_of_source_columns": 3}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["recurrence.cached_end_date_tz"], "Num_of_source_columns": 1}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_I5_Calendar_20260127T183815Z.jsonl
View File

@@ -1,5 +0,0 @@
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "EMAIL", "PII": ["ottomatik1234@gmail.com"], "Num_of_PII": 1, "source_columns": ["alarm.email_address", "calendar.shared_owner_address", "calendar.self_identity_email", "calendar.owner_identity_email"], "Num_of_source_columns": 4}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "PERSON_NAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

5
RQs/normalized_PII_results/ground_truth/db_level/PII_I5_Extras.jsonl Normal file
View File

@@ -0,0 +1,5 @@
{"db_path": "selectedDBs\\I5_Extras.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I5_Extras.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I5_Extras.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I5_Extras.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}
{"db_path": "selectedDBs\\I5_Extras.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

BIN
RQs/usenixsecurity2026.pdf Normal file
View File

Binary file not shown.