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

change name to match models

Browse Source
This commit is contained in:
Frank Xu
2026-01-28 20:16:13 -05:00
parent 05bfa15e85
commit 8e6158bfd7
51 changed files with 100 additions and 70 deletions
Download Patch File Download Diff File Expand all files Collapse all files

31
RQs/RQ0/RQ0_batch_results_normalization.ipynb → RQs/RQ0/RQ0_results_normalization.ipynb
View File

@@ -36,7 +36,34 @@
"PII_I5_Calendar_20260127T183815Z.jsonl: 5 records -> ..\\batch_results_normalized\\PII_I5_Calendar_20260127T183815Z.jsonl\n",
"PII_I5_Extras_20260127T183857Z.jsonl: 5 records -> ..\\batch_results_normalized\\PII_I5_Extras_20260127T183857Z.jsonl\n",
"Done. Files: 25, Records: 125\n",
"Output folder: I:\\project2026\\llmagent\\RQs\\batch_results_normalized\n"
"Output folder: I:\\project2026\\llmagent\\RQs\\batch_results_normalized\n",
"PII_A1_commerce.jsonl: 1 records -> ..\\ground_truth_normalized\\PII_A1_commerce.jsonl\n",
"PII_A1_msgstore.jsonl: 1 records -> ..\\ground_truth_normalized\\PII_A1_msgstore.jsonl\n",
"PII_A1_wa.jsonl: 8 records -> ..\\ground_truth_normalized\\PII_A1_wa.jsonl\n",
"PII_A2_core.jsonl: 4 records -> ..\\ground_truth_normalized\\PII_A2_core.jsonl\n",
"PII_A2_journal.jsonl: 1 records -> ..\\ground_truth_normalized\\PII_A2_journal.jsonl\n",
"PII_A2_main.jsonl: 10 records -> ..\\ground_truth_normalized\\PII_A2_main.jsonl\n",
"PII_A3_account1cache4.jsonl: 1 records -> ..\\ground_truth_normalized\\PII_A3_account1cache4.jsonl\n",
"PII_A3_account2cache4.jsonl: 1 records -> ..\\ground_truth_normalized\\PII_A3_account2cache4.jsonl\n",
"PII_A3_account3cache4.jsonl: 1 records -> ..\\ground_truth_normalized\\PII_A3_account3cache4.jsonl\n",
"PII_A4_gmm_myplaces.jsonl: 1 records -> ..\\ground_truth_normalized\\PII_A4_gmm_myplaces.jsonl\n",
"PII_A4_gmm_storage.jsonl: 1 records -> ..\\ground_truth_normalized\\PII_A4_gmm_storage.jsonl\n",
"PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14.jsonl: 6 records -> ..\\ground_truth_normalized\\PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14.jsonl\n",
"PII_A5_SBrowser.jsonl: 3 records -> ..\\ground_truth_normalized\\PII_A5_SBrowser.jsonl\n",
"PII_A5_SBrowser2.jsonl: 1 records -> ..\\ground_truth_normalized\\PII_A5_SBrowser2.jsonl\n",
"PII_A5_searchengine.jsonl: 1 records -> ..\\ground_truth_normalized\\PII_A5_searchengine.jsonl\n",
"PII_I1_CallHistory.jsonl: 1 records -> ..\\ground_truth_normalized\\PII_I1_CallHistory.jsonl\n",
"PII_I1_ChatStorage.jsonl: 4 records -> ..\\ground_truth_normalized\\PII_I1_ChatStorage.jsonl\n",
"PII_I1_ContactsV2.jsonl: 2 records -> ..\\ground_truth_normalized\\PII_I1_ContactsV2.jsonl\n",
"PII_I2_AddressBook.jsonl: 7 records -> ..\\ground_truth_normalized\\PII_I2_AddressBook.jsonl\n",
"PII_I2_AddressBookImages.jsonl: 1 records -> ..\\ground_truth_normalized\\PII_I2_AddressBookImages.jsonl\n",
"PII_I3_sms.jsonl: 6 records -> ..\\ground_truth_normalized\\PII_I3_sms.jsonl\n",
"PII_I4_CloudTabs.jsonl: 1 records -> ..\\ground_truth_normalized\\PII_I4_CloudTabs.jsonl\n",
"PII_I4_History.jsonl: 1 records -> ..\\ground_truth_normalized\\PII_I4_History.jsonl\n",
"PII_I5_Calendar.jsonl: 6 records -> ..\\ground_truth_normalized\\PII_I5_Calendar.jsonl\n",
"PII_I5_Extras.jsonl: 1 records -> ..\\ground_truth_normalized\\PII_I5_Extras.jsonl\n",
"Done. Files: 25, Records: 71\n",
"Output folder: I:\\project2026\\llmagent\\RQs\\ground_truth_normalized\n"
]
}
],
@@ -113,7 +140,7 @@
" from stats_utils import normalize_and_slim_record\n",
"\n",
" IN_DIR = Path(r\"..\\..\\batch_results\")\n",
" OUT_DIR = Path(r\"..\\batch_results_normalized\")\n",
" OUT_DIR = Path(r\"..\\batch_results_gpt4o_normalized\")\n",
"\n",
" normalize_jsonl_folder(IN_DIR, OUT_DIR, normalize_and_slim_record, delete_out_dir_first=True)\n",
" \n",

0
RQs/batch_results_normalized/PII_A1_commerce_20260127T175911Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_A1_commerce_20260127T175911Z.jsonl
View File

0
RQs/batch_results_normalized/PII_A1_msgstore_20260127T180043Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_A1_msgstore_20260127T180043Z.jsonl
View File

0
RQs/batch_results_normalized/PII_A1_wa_20260127T180213Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_A1_wa_20260127T180213Z.jsonl
View File

0
RQs/batch_results_normalized/PII_A2_core_20260127T180339Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_A2_core_20260127T180339Z.jsonl
View File

0
RQs/batch_results_normalized/PII_A2_journal_20260127T180440Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_A2_journal_20260127T180440Z.jsonl
View File

0
RQs/batch_results_normalized/PII_A2_main_20260127T180710Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_A2_main_20260127T180710Z.jsonl
View File

0
RQs/batch_results_normalized/PII_A3_account1cache4_20260127T180745Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_A3_account1cache4_20260127T180745Z.jsonl
View File

0
RQs/batch_results_normalized/PII_A3_account2cache4_20260127T180821Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_A3_account2cache4_20260127T180821Z.jsonl
View File

0
RQs/batch_results_normalized/PII_A3_account3cache4_20260127T180857Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_A3_account3cache4_20260127T180857Z.jsonl
View File

0
RQs/batch_results_normalized/PII_A4_gmm_myplaces_20260127T180935Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_A4_gmm_myplaces_20260127T180935Z.jsonl
View File

0
RQs/batch_results_normalized/PII_A4_gmm_storage_20260127T181014Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_A4_gmm_storage_20260127T181014Z.jsonl
View File

0
RQs/batch_results_normalized/PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14_20260127T181121Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14_20260127T181121Z.jsonl
View File

0
RQs/batch_results_normalized/PII_A5_SBrowser2_20260127T181345Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_A5_SBrowser2_20260127T181345Z.jsonl
View File

0
RQs/batch_results_normalized/PII_A5_SBrowser_20260127T181239Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_A5_SBrowser_20260127T181239Z.jsonl
View File

0
RQs/batch_results_normalized/PII_A5_searchengine_20260127T181446Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_A5_searchengine_20260127T181446Z.jsonl
View File

0
RQs/batch_results_normalized/PII_I1_CallHistory_20260127T181557Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_I1_CallHistory_20260127T181557Z.jsonl
View File

0
RQs/batch_results_normalized/PII_I1_ChatStorage_20260127T181731Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_I1_ChatStorage_20260127T181731Z.jsonl
View File

0
RQs/batch_results_normalized/PII_I1_ContactsV2_20260127T182906Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_I1_ContactsV2_20260127T182906Z.jsonl
View File

0
RQs/batch_results_normalized/PII_I2_AddressBookImages_20260127T183526Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_I2_AddressBookImages_20260127T183526Z.jsonl
View File

0
RQs/batch_results_normalized/PII_I2_AddressBook_20260127T183457Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_I2_AddressBook_20260127T183457Z.jsonl
View File

0
RQs/batch_results_normalized/PII_I3_sms_20260127T183606Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_I3_sms_20260127T183606Z.jsonl
View File

0
RQs/batch_results_normalized/PII_I4_CloudTabs_20260127T183643Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_I4_CloudTabs_20260127T183643Z.jsonl
View File

0
RQs/batch_results_normalized/PII_I4_History_20260127T183727Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_I4_History_20260127T183727Z.jsonl
View File

0
RQs/batch_results_normalized/PII_I5_Calendar_20260127T183815Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_I5_Calendar_20260127T183815Z.jsonl
View File

0
RQs/batch_results_normalized/PII_I5_Extras_20260127T183857Z.jsonl → RQs/batch_results_gpt4o_normalized/PII_I5_Extras_20260127T183857Z.jsonl
View File

2
RQs/ground_truth_normalized/PII_A1_commerce.jsonl
View File

@@ -1 +1 @@
{"db_path": "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}

2
RQs/ground_truth_normalized/PII_A1_msgstore.jsonl
View File

@@ -1 +1 @@
{"db_path": "msgstore.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["business_name"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A1_msgstore.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["business_name"], "Num_of_source_columns": 1}

16
RQs/ground_truth_normalized/PII_A1_wa.jsonl
View File

@@ -1,8 +1,8 @@
{"db_path": "wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["number"], "Num_of_source_columns": 1}
{"db_path": "wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["given_name+family_name", "sort_name"], "Num_of_source_columns": 2}
{"db_path": "wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["display_name"], "Num_of_source_columns": 1}
{"db_path": "wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["number"], "Num_of_source_columns": 1}
{"db_path": "wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["display_name", "wa_name"], "Num_of_source_columns": 2}
{"db_path": "wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["given_name+family_name", "sort_name"], "Num_of_source_columns": 2}
{"db_path": "wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["verified_name"], "Num_of_source_columns": 1}
{"db_path": "wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["creator_name"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["number"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["given_name+family_name", "sort_name"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["display_name"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["number"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["display_name", "wa_name"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["given_name+family_name", "sort_name"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["verified_name"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A1_wa.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["creator_name"], "Num_of_source_columns": 1}

8
RQs/ground_truth_normalized/PII_A2_core.jsonl
View File

@@ -1,4 +1,4 @@
{"db_path": "core.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["textval"], "Num_of_source_columns": 1}
{"db_path": "core.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["textval"], "Num_of_source_columns": 1}
{"db_path": "core.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["textval"], "Num_of_source_columns": 1}
{"db_path": "core.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["textval"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["textval"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["textval"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["textval"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A2_core.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["textval"], "Num_of_source_columns": 1}

2
RQs/ground_truth_normalized/PII_A2_journal.jsonl
View File

@@ -1 +1 @@
{"db_path": "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}

20
RQs/ground_truth_normalized/PII_A2_main.jsonl
View File

@@ -1,10 +1,10 @@
{"db_path": "main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["encodedusername", "mutableusername", "originalusername"], "Num_of_source_columns": 3}
{"db_path": "main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["phone", "rawphone"], "Num_of_source_columns": 2}
{"db_path": "main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["displayname"], "Num_of_source_columns": 1}
{"db_path": "main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["displayname", "serverdisplayname", "username"], "Num_of_source_columns": 3}
{"db_path": "main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["displayname"], "Num_of_source_columns": 1}
{"db_path": "main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["captiontextdisplay"], "Num_of_source_columns": 1}
{"db_path": "main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["displayname", "serverdisplayname", "username", "usernameforsorting"], "Num_of_source_columns": 4}
{"db_path": "main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["displayinteractionuserusername", "frienddisplayusername"], "Num_of_source_columns": 2}
{"db_path": "main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["displayinteractionuserdisplayname"], "Num_of_source_columns": 1}
{"db_path": "main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["frienddisplayname", "friendusername"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["encodedusername", "mutableusername", "originalusername"], "Num_of_source_columns": 3}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["phone", "rawphone"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["displayname"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["displayname", "serverdisplayname", "username"], "Num_of_source_columns": 3}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["displayname"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["captiontextdisplay"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["displayname", "serverdisplayname", "username", "usernameforsorting"], "Num_of_source_columns": 4}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["displayinteractionuserusername", "frienddisplayusername"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["displayinteractionuserdisplayname"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A2_main.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["frienddisplayname", "friendusername"], "Num_of_source_columns": 2}

2
RQs/ground_truth_normalized/PII_A3_account1cache4.jsonl
View File

@@ -1 +1 @@
{"db_path": "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}

2
RQs/ground_truth_normalized/PII_A3_account2cache4.jsonl
View File

@@ -1 +1 @@
{"db_path": "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}

2
RQs/ground_truth_normalized/PII_A3_account3cache4.jsonl
View File

@@ -1 +1 @@
{"db_path": "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}

2
RQs/ground_truth_normalized/PII_A4_gmm_myplaces.jsonl
View File

@@ -1 +1 @@
{"db_path": "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}

2
RQs/ground_truth_normalized/PII_A4_gmm_storage.jsonl
View File

@@ -1 +1 @@
{"db_path": "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}

12
RQs/ground_truth_normalized/PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14.jsonl
View File

@@ -1,6 +1,6 @@
{"db_path": "peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["value"], "Num_of_source_columns": 1}
{"db_path": "peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["value"], "Num_of_source_columns": 1}
{"db_path": "peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["value"], "Num_of_source_columns": 1}
{"db_path": "peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["c1value"], "Num_of_source_columns": 1}
{"db_path": "peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["c1value"], "Num_of_source_columns": 1}
{"db_path": "peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["c1value"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["value"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["value"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["value"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["c1value"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["c1value"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\A4_peopleCache_sharononeil368@gmail.com_com.google_14.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["c1value"], "Num_of_source_columns": 1}

6
RQs/ground_truth_normalized/PII_A5_SBrowser.jsonl
View File

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

2
RQs/ground_truth_normalized/PII_A5_SBrowser2.jsonl
View File

@@ -1 +1 @@
{"db_path": "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}

2
RQs/ground_truth_normalized/PII_A5_searchengine.jsonl
View File

@@ -1 +1 @@
{"db_path": "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}

2
RQs/ground_truth_normalized/PII_I1_CallHistory.jsonl
View File

@@ -1 +1 @@
{"db_path": "CallHistory.sqlitedb", "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}

8
RQs/ground_truth_normalized/PII_I1_ChatStorage.jsonl
View File

@@ -1,4 +1,4 @@
{"db_path": "ChatStorage.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zpartnername"], "Num_of_source_columns": 1}
{"db_path": "ChatStorage.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zpartnername"], "Num_of_source_columns": 1}
{"db_path": "ChatStorage.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zpushname"], "Num_of_source_columns": 1}
{"db_path": "ChatStorage.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zpushname"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zpartnername"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zpartnername"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zpushname"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zpushname"], "Num_of_source_columns": 1}

4
RQs/ground_truth_normalized/PII_I1_ContactsV2.jsonl
View File

@@ -1,2 +1,2 @@
{"db_path": "ContactsV2.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zfullname"], "Num_of_source_columns": 1}
{"db_path": "ContactsV2.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zphonenumber"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I1_ContactsV2.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zfullname"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I1_ContactsV2.sqlite", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["zphonenumber"], "Num_of_source_columns": 1}

12
RQs/ground_truth_normalized/PII_I2_AddressBook.jsonl
View File

@@ -1,5 +1,7 @@
{"db_path": "AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["first+last"], "Num_of_source_columns": 1}
{"db_path": "AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["value"], "Num_of_source_columns": 1}
{"db_path": "AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["value"], "Num_of_source_columns": 1}
{"db_path": "AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["c17email"], "Num_of_source_columns": 1}
{"db_path": "AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["c0first+c1last+c2middle"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I2_AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["first+last"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I2_AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["value"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I2_AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["value"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I2_AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["c17email"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I2_AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["c0first+c1last+c2middle"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I2_AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["value"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I2_AddressBook.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["c18address"], "Num_of_source_columns": 1}

2
RQs/ground_truth_normalized/PII_I2_AddressBookImages.jsonl
View File

@@ -1 +1 @@
{"db_path": "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}

12
RQs/ground_truth_normalized/PII_I3_sms.jsonl
View File

@@ -1,6 +1,6 @@
{"db_path": "sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["account_login", "chat_identifier", "guid", "last_addressed_handle"], "Num_of_source_columns": 4}
{"db_path": "sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["account_login", "chat_identifier", "guid"], "Num_of_source_columns": 3}
{"db_path": "sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["account", "destination_caller_id"], "Num_of_source_columns": 2}
{"db_path": "sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["account"], "Num_of_source_columns": 1}
{"db_path": "sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["id", "uncanonicalized_id"], "Num_of_source_columns": 2}
{"db_path": "sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["id"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["account_login", "chat_identifier", "guid", "last_addressed_handle"], "Num_of_source_columns": 4}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["account_login", "chat_identifier", "guid"], "Num_of_source_columns": 3}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["account", "destination_caller_id"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["account"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["id", "uncanonicalized_id"], "Num_of_source_columns": 2}
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["id"], "Num_of_source_columns": 1}

2
RQs/ground_truth_normalized/PII_I4_CloudTabs.jsonl
View File

@@ -1 +1 @@
{"db_path": "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}

2
RQs/ground_truth_normalized/PII_I4_History.jsonl
View File

@@ -1 +1 @@
{"db_path": "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}

11
RQs/ground_truth_normalized/PII_I5_Calendar.jsonl
View File

@@ -1,5 +1,6 @@
{"db_path": "Calendar.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["last_sync_title", "notes", "owner_identity_email", "self_identity_email", "shared_owner_address", "title"], "Num_of_source_columns": 6}
{"db_path": "Calendar.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["shared_owner_name"], "Num_of_source_columns": 1}
{"db_path": "Calendar.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["address"], "Num_of_source_columns": 1}
{"db_path": "Calendar.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["display_name"], "Num_of_source_columns": 1}
{"db_path": "Calendar.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["owner_name"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["last_sync_title", "notes", "owner_identity_email", "self_identity_email", "shared_owner_address", "title"], "Num_of_source_columns": 6}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["shared_owner_name"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["address"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["display_name"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["owner_name"], "Num_of_source_columns": 1}
{"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "", "PII": [], "Num_of_PII": 0, "source_columns": ["cached_end_date_tz"], "Num_of_source_columns": 1}

2
RQs/ground_truth_normalized/PII_I5_Extras.jsonl
View File

@@ -1 +1 @@
{"db_path": "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}