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

update RQs

Browse Source
This commit is contained in:
Frank Xu
2026-02-18 16:22:33 -05:00
parent 537cdad635
commit efd042b0e0
38 changed files with 716 additions and 151 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
RQs/RQ0/RQ0_1_results_normalization.ipynb
View File

@@ -2,7 +2,7 @@
"cells": [ "cells": [
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 3, "execution_count": 1,
"id": "c2d824a6", "id": "c2d824a6",
"metadata": {}, "metadata": {},
"outputs": [ "outputs": [

2
RQs/RQ0/RQ0_2_app_level_aggregation.ipynb
View File

@@ -2,7 +2,7 @@
"cells": [ "cells": [
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 2, "execution_count": 1,
"id": "234eed3f", "id": "234eed3f",
"metadata": {}, "metadata": {},
"outputs": [ "outputs": [

2
RQs/RQ0/RQ0_3_corpus_level_aggreation.ipynb
View File

@@ -2,7 +2,7 @@
"cells": [ "cells": [
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 3, "execution_count": 1,
"id": "234eed3f", "id": "234eed3f",
"metadata": {}, "metadata": {},
"outputs": [ "outputs": [

31
RQs/RQ2/RQ2_search_space_reduction.tex Normal file
View File

@@ -0,0 +1,31 @@
\begin{table}[th]
\centering
\caption{Reduction of effective extraction space via hypothesis-driven planning. Total candidate columns are counted over all tables in the selected databases for each application, while reductions correspond to columns exhaustively scanned during row-level PII extraction.}
\label{tab:search_space_reduction}
\small
\begin{tabular}{|l|l|p{1.3cm}|p{1.7cm}|p{1.0cm}|}
\hline
\textbf{ID} & \textbf{Apps} & \textbf{Candidate Cols (Total)} & \textbf{Cols Scanned (Extraction)} & \textbf{Reduc. (\%)} \\
\hline
A1 & WhatsApp & 1627 & 40 & 97.54\% \\
\hline
A2 & Snapchat & 842 & 161 & 80.88\% \\
\hline
A3 & Telegram & 1197 & 0 & 100.00\% \\
\hline
A4 & Google Maps & 71 & 2 & 97.18\% \\
\hline
A5 & Samsung Internet & 173 & 43 & 75.14\% \\
\hline
I1 & WhatsApp & 328 & 44 & 86.59\% \\
\hline
I2 & Contacts & 219 & 17 & 92.24\% \\
\hline
I3 & Apple Messages & 181 & 39 & 78.45\% \\
\hline
I4 & Safari & 72 & 0 & 100.00\% \\
\hline
I5 & Calendar & 539 & 35 & 93.51\% \\
\hline
\end{tabular}
\end{table}

4
RQs/RQ2/RQ2_search_space_reduction_gpt4o.tex
View File

@@ -27,9 +27,5 @@ I4 & Safari & 74 & 0 & 100.00\% \\
\hline \hline
I5 & Calendar & 541 & 0 & 100.00\% \\ I5 & Calendar & 541 & 0 & 100.00\% \\
\hline \hline
test2 & test2 & 12 & 0 & 100.00\% \\
\hline
users & users & 3 & 0 & 100.00\% \\
\hline
\end{tabular} \end{tabular}
\end{table} \end{table}

6
RQs/RQ2/RQ2_t6_0_total_columns.ipynb
View File

@@ -2,7 +2,7 @@
"cells": [ "cells": [
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 2, "execution_count": 1,
"id": "2c488f9e", "id": "2c488f9e",
"metadata": {}, "metadata": {},
"outputs": [ "outputs": [
@@ -10,7 +10,7 @@
"name": "stdout", "name": "stdout",
"output_type": "stream", "output_type": "stream",
"text": [ "text": [
"Wrote: C:\\Users\\cyfij\\OneDrive\\Desktop\\DFRWS 2026\\Agent\\RQs\\RQ2\\app_total_columns.csv\n" "Wrote: I:\\project2026\\llmagent\\RQs\\RQ2\\app_total_columns.csv\n"
] ]
} }
], ],
@@ -188,7 +188,7 @@
"name": "python", "name": "python",
"nbconvert_exporter": "python", "nbconvert_exporter": "python",
"pygments_lexer": "ipython3", "pygments_lexer": "ipython3",
"version": "3.14.0" "version": "3.10.18"
} }
}, },
"nbformat": 4, "nbformat": 4,

18
RQs/RQ2/app_total_columns.csv
View File

@@ -1,13 +1,11 @@
app_code,app_name,total_columns app_code,app_name,total_columns
A1,WhatsApp,1637 A1,WhatsApp,1627
A2,Snapchat,848 A2,Snapchat,842
A3,Telegram,1197 A3,Telegram,1197
A4,Google Maps,80 A4,Google Maps,71
A5,Samsung Internet,185 A5,Samsung Internet,173
I1,WhatsApp,328 I1,WhatsApp,328
I2,Contacts,13 I2,Contacts,219
I3,Apple Messages,186 I3,Apple Messages,181
I4,Safari,74 I4,Safari,72
I5,Calendar,541 I5,Calendar,539
test2,test2,12
users,users,3
1 app_code app_name total_columns
2 A1 WhatsApp 1637 1627
3 A2 Snapchat 848 842
4 A3 Telegram 1197
5 A4 Google Maps 80 71
6 A5 Samsung Internet 185 173
7 I1 WhatsApp 328
8 I2 Contacts 13 219
9 I3 Apple Messages 186 181
10 I4 Safari 74 72
11 I5 Calendar 541 539
test2 test2 12
users users 3

12
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": null, "execution_count": 1,
"id": "234eed3f", "id": "234eed3f",
"metadata": {}, "metadata": {},
"outputs": [ "outputs": [
@@ -17,15 +17,15 @@
"\\textbf{Recall} &\n", "\\textbf{Recall} &\n",
"\\textbf{Precision} \\\\\n", "\\textbf{Precision} \\\\\n",
"\\hline\n", "\\hline\n",
"Email Address & 10 & 10 & 10 & 100.0\\% & 100.0\\% \\\\\n", "Email Address & 15 & 16 & 15 & 100.0\\% & 93.8\\% \\\\\n",
"\\hline\n", "\\hline\n",
"Phone Number & 1050 & 801 & 225 & 21.4\\% & 28.1\\% \\\\\n", "Phone Number & 1051 & 1184 & 734 & 69.8\\% & 62.0\\% \\\\\n",
"\\hline\n", "\\hline\n",
"User Name & 85 & 751 & 8 & 9.4\\% & 1.1\\% \\\\\n", "User Name & 3446 & 1234 & 1232 & 35.8\\% & 99.8\\% \\\\\n",
"\\hline\n", "\\hline\n",
"Person Name & 909 & 787 & 783 & 86.1\\% & 99.5\\% \\\\\n", "Person Name & 1134 & 2157 & 1087 & 95.9\\% & 50.4\\% \\\\\n",
"\\hline\n", "\\hline\n",
"Postal Address & 2 & 2 & 2 & 100.0\\% & 100.0\\% \\\\\n", "Postal Address & 21 & 2 & 2 & 9.5\\% & 100.0\\% \\\\\n",
"\\hline\n" "\\hline\n"
] ]
} }

232
RQs/RQ4/RQ4_t11.ipynb Normal file
View File

@@ -0,0 +1,232 @@
{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"id": "ebe9dbb1",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\\begin{table*}\n",
"\\centering\n",
"\\small\n",
"\\caption{Distribution of distinct PII entities discovered across evaluated methods and PII categories.}\n",
"\\label{tab:model_yield}\n",
"\\begin{tabular}{|l|l|l|l|l|l|l|l|}\n",
"\\hline\n",
"\\multicolumn{2}{|c|}{\\textbf{Method/LLM}} & \\textbf{Email} & \\textbf{Phone} & \\textbf{User Name} & \\textbf{Real Name} & \\textbf{Precision} & \\textbf{Recall} \\\\\n",
"\\hline\n",
"Gemini-2.5-Pro & 88.4\\% & 10 & 791 & 1664 & 1076 & 76.50\\% & 49.03\\% \\\\\n",
"\\hline\n",
"GPT-3.5-Turbo & xx & 4 & 13 & 0 & 276 & 24.23\\% & 1.29\\% \\\\\n",
"\\hline\n",
"GPT-4.1 & xx & 1289 & 680 & 531 & 683 & 33.30\\% & 19.19\\% \\\\\n",
"\\hline\n",
"GPT-4o-mini & 82.0\\% & 14 & 22 & 875 & 1928 & 39.98\\% & 20.54\\% \\\\\n",
"\\hline\n",
"GPT-5.1 & xx & 16 & 1184 & 1234 & 2154 & 51.20\\% & 42.52\\% \\\\\n",
"\\hline\n",
"LLaMA-3.1-8B-Instruct & xx & 0 & 0 & 2 & 15 & 88.24\\% & 0.27\\% \\\\\n",
"\\hline\n",
"LLaMA-3.1-70B-Instruct & xx & 6 & 6 & 34 & 15 & 39.34\\% & 0.43\\% \\\\\n",
"\\hline\n",
"Mistral-Large & xx & 2 & 989 & 2121 & 1 & 59.56\\% & 33.56\\% \\\\\n",
"\\hline\n",
"Mixtral-8x7B & xx & 6 & 6 & 2302 & 98 & 48.34\\% & 21.10\\% \\\\\n",
"\\hline\n",
"Mixtral-8x22B & xx & 0 & 6 & 58 & 15 & 25.32\\% & 0.36\\% \\\\\n",
"\\hline\n",
"Qwen2.5-72B & 82.3\\% & 961 & 45 & 528 & 1203 & 38.87\\% & 19.26\\% \\\\\n",
"\\hline\n",
"\\end{tabular}\n",
"\\end{table*}\n"
]
}
],
"source": [
"import json\n",
"import re\n",
"from pathlib import Path\n",
"from typing import Dict, Set\n",
"\n",
"# =========================\n",
"# CONFIG\n",
"# =========================\n",
"\n",
"BASE_DIR = Path(r\"C:\\Users\\cyfij\\OneDrive\\Desktop\\DFRWS 2026\\Agent\\RQs\\normalized_results\")\n",
"\n",
"GT_PATH = BASE_DIR / \"ground_truth\" / \"corpus_level\" / \"corpus_level.jsonl\"\n",
"\n",
"METHODS = {\n",
" \"Gemini-2.5-Pro\": {\"path\": \"gemini_2_5_pro\", \"bm\": \"88.4\\\\%\"},\n",
" \"GPT-3.5-Turbo\": {\"path\": \"gpt_3_5_turbo\", \"bm\": \"xx\"},\n",
" \"GPT-4.1\": {\"path\": \"gpt_4_1\", \"bm\": \"xx\"},\n",
" \"GPT-4o-mini\": {\"path\": \"gpt_4o_mini\", \"bm\": \"82.0\\\\%\"},\n",
" \"GPT-5.1\": {\"path\": \"gpt_5_1\", \"bm\": \"xx\"},\n",
" \"LLaMA-3.1-8B-Instruct\": {\"path\": \"llama_3_1_8b\", \"bm\": \"xx\"},\n",
" \"LLaMA-3.1-70B-Instruct\": {\"path\": \"llama_3_1_70b\", \"bm\": \"xx\"},\n",
" \"Mistral-Large\": {\"path\": \"mistral_large\", \"bm\": \"xx\"},\n",
" \"Mixtral-8x7B\": {\"path\": \"mixtral_8x7b\", \"bm\": \"xx\"},\n",
" \"Mixtral-8x22B\": {\"path\": \"mixtral_8x22b\", \"bm\": \"xx\"},\n",
" \"Qwen2.5-72B\": {\"path\": \"qwen_2_5_72b\", \"bm\": \"82.3\\\\%\"},\n",
"}\n",
"\n",
"PII_ORDER = [\"EMAIL\", \"PHONE\", \"USERNAME\", \"PERSON_NAME\"]\n",
"\n",
"# =========================\n",
"# CANONICALIZATION\n",
"# =========================\n",
"\n",
"def canonicalize(val: str, pii_type: str) -> str:\n",
" val = val.strip()\n",
"\n",
" if pii_type == \"EMAIL\":\n",
" return val.lower()\n",
"\n",
" if pii_type == \"PHONE\":\n",
" plus = val.startswith(\"+\")\n",
" digits = re.sub(r\"\\D\", \"\", val)\n",
" return \"+\" + digits if plus else digits\n",
"\n",
" if pii_type in {\"USERNAME\", \"PERSON_NAME\"}:\n",
" val = val.lower()\n",
" val = re.sub(r\"\\b(mr|ms|mrs|dr|prof)\\.?\\b\", \"\", val)\n",
" val = re.sub(r\"[^\\w\\s]\", \"\", val)\n",
" val = re.sub(r\"\\s+\", \" \", val)\n",
" return val.strip()\n",
"\n",
" return val\n",
"\n",
"\n",
"# =========================\n",
"# LOAD CORPUS\n",
"# =========================\n",
"\n",
"def load_corpus(path: Path) -> Dict[str, Set[str]]:\n",
" data = {t: set() for t in PII_ORDER}\n",
" if not path.exists():\n",
" return data\n",
"\n",
" with path.open(\"r\", encoding=\"utf-8\") as f:\n",
" for line in f:\n",
" rec = json.loads(line)\n",
" pii_type = (rec.get(\"PII_type\") or \"\").upper()\n",
" if pii_type not in PII_ORDER:\n",
" continue\n",
"\n",
" vals = rec.get(\"PII_unique\") or rec.get(\"PII_all\") or []\n",
" canon_vals = {\n",
" canonicalize(v, pii_type)\n",
" for v in vals\n",
" if isinstance(v, str) and v.strip()\n",
" }\n",
"\n",
" data[pii_type].update(canon_vals)\n",
"\n",
" return data\n",
"\n",
"# =========================\n",
"# LOAD GT\n",
"# =========================\n",
"\n",
"GT = load_corpus(GT_PATH)\n",
"\n",
"# =========================\n",
"# COMPUTE TABLE\n",
"# =========================\n",
"\n",
"rows = []\n",
"\n",
"for name, info in METHODS.items():\n",
"\n",
" corpus_path = BASE_DIR / info[\"path\"] / \"corpus_level\" / \"corpus_level.jsonl\"\n",
" SYS = load_corpus(corpus_path)\n",
"\n",
" total_sys = 0\n",
" total_gt = 0\n",
" total_overlap = 0\n",
" counts = {}\n",
"\n",
" for t in PII_ORDER:\n",
"\n",
" gt_set = GT[t]\n",
" sys_set = SYS[t]\n",
"\n",
" counts[t] = len(sys_set)\n",
"\n",
" overlap = gt_set.intersection(sys_set)\n",
"\n",
" total_sys += len(sys_set)\n",
" total_gt += len(gt_set)\n",
" total_overlap += len(overlap)\n",
"\n",
" precision = (total_overlap / total_sys) if total_sys else 0.0\n",
" recall = (total_overlap / total_gt) if total_gt else 0.0\n",
"\n",
" rows.append((\n",
" name,\n",
" info[\"bm\"],\n",
" counts[\"EMAIL\"],\n",
" counts[\"PHONE\"],\n",
" counts[\"USERNAME\"],\n",
" counts[\"PERSON_NAME\"],\n",
" f\"{precision*100:.2f}\\\\%\",\n",
" f\"{recall*100:.2f}\\\\%\",\n",
" ))\n",
"\n",
"# =========================\n",
"# EMIT LATEX\n",
"# =========================\n",
"\n",
"print(r\"\\begin{table*}\")\n",
"print(r\"\\centering\")\n",
"print(r\"\\small\")\n",
"print(r\"\\caption{Distribution of distinct PII entities discovered across evaluated methods and PII categories.}\")\n",
"print(r\"\\label{tab:model_yield}\")\n",
"print(r\"\\begin{tabular}{|l|l|l|l|l|l|l|l|}\")\n",
"print(r\"\\hline\")\n",
"print(r\"\\multicolumn{2}{|c|}{\\textbf{Method/LLM}} & \\textbf{Email} & \\textbf{Phone} & \\textbf{User Name} & \\textbf{Real Name} & \\textbf{Precision} & \\textbf{Recall} \\\\\")\n",
"print(r\"\\hline\")\n",
"\n",
"for r in rows:\n",
" print(\" & \".join(map(str, r)) + r\" \\\\\")\n",
" print(r\"\\hline\")\n",
"\n",
"print(r\"\\end{tabular}\")\n",
"print(r\"\\end{table*}\")\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "07108960",
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.14.0"
}
},
"nbformat": 4,
"nbformat_minor": 5
}

253
RQs/RQ4/RQ4_t12.ipynb Normal file
View File

@@ -0,0 +1,253 @@
{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"id": "03fbbc9f",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"CSV path: C:\\Users\\cyfij\\OneDrive\\Desktop\\DFRWS 2026\\Agent\\RQs\\RQ2\\app_total_columns.csv\n",
"Exists: True\n",
"Using BASE_DIR: C:\\Users\\cyfij\\OneDrive\\Desktop\\DFRWS 2026\\Agent\\RQs\n",
"Using CSV: C:\\Users\\cyfij\\OneDrive\\Desktop\\DFRWS 2026\\Agent\\RQs\\RQ2\\app_total_columns.csv\n",
"CSV exists: True\n",
"\\begin{tabular}{|l|p{1.4cm}|p{1.8cm}|}\n",
"\\hline\n",
"\\textbf{Method/LLM} &\n",
"\\textbf{Avg. Cols Examined} &\n",
"\\textbf{Avg. Search Space Reduc.} \\\\\n",
"\\hline\n",
"bulk\\_extractor-v1.6 (baseline) & NA & 0.0\\% \\\\\n",
"\\hline\n",
"GPT-4o-mini & 9.1 & 85.0\\% \\\\\n",
"\\hline\n",
"Gemini-2.5-Pro & 0 & 100.0\\% \\\\\n",
"\\hline\n",
"Qwen2.5-72B & 31.5 & 41.2\\% \\\\\n",
"\\hline\n",
"LLaMA-3.1-70B-Instruct & 1 & 97.5\\% \\\\\n",
"\\hline\n",
"LLaMA-3.1-8B-Instruct & 0.4 & 99.9\\% \\\\\n",
"\\hline\n",
"Mixtral-8x22B & 3.2 & 90.3\\% \\\\\n",
"\\hline\n",
"Mixtral-8x7B & 2.6 & 97.2\\% \\\\\n",
"\\hline\n",
"Mistral-Large & 24.9 & 75.6\\% \\\\\n",
"\\hline\n",
"GPT-5.1 & 29.9 & 81.0\\% \\\\\n",
"\\hline\n",
"GPT-4.1 & 20.5 & 69.7\\% \\\\\n",
"\\hline\n",
"GPT-3.5-turbo & 2.8 & 98.6\\% \\\\\n",
"\\hline\n",
"\\end{tabular}\n"
]
}
],
"source": [
"import json\n",
"import csv\n",
"from pathlib import Path\n",
"from statistics import mean\n",
"\n",
"# -------------------------------------------------\n",
"# Auto-detect project root (Agent directory)\n",
"# -------------------------------------------------\n",
"BASE_DIR = Path(r\"C:\\Users\\cyfij\\OneDrive\\Desktop\\DFRWS 2026\\Agent\\RQs\")\n",
"\n",
"\n",
"CANDIDATE_CSV = BASE_DIR / \"RQ2\" / \"app_total_columns.csv\"\n",
"\n",
"print(\"CSV path:\", CANDIDATE_CSV)\n",
"print(\"Exists:\", CANDIDATE_CSV.exists())\n",
"\n",
"\n",
"print(\"Using BASE_DIR:\", BASE_DIR)\n",
"print(\"Using CSV:\", CANDIDATE_CSV)\n",
"print(\"CSV exists:\", CANDIDATE_CSV.exists())\n",
"\n",
"# -------------------------------------------------\n",
"# MODELS\n",
"# -------------------------------------------------\n",
"MODELS = {\n",
" \"gpt_4o_mini\": \"GPT-4o-mini\",\n",
" \"gemini_2_5_pro\": \"Gemini-2.5-Pro\",\n",
" \"qwen_2_5_72b\": \"Qwen2.5-72B\",\n",
" \"llama_3_1_70b\": \"LLaMA-3.1-70B-Instruct\",\n",
" \"llama_3_1_8b\": \"LLaMA-3.1-8B-Instruct\",\n",
" \"mixtral_8x22b\": \"Mixtral-8x22B\",\n",
" \"mixtral_8x7b\": \"Mixtral-8x7B\",\n",
" \"mistral_large\": \"Mistral-Large\",\n",
" \"gpt_5_1\": \"GPT-5.1\",\n",
" \"gpt_4_1\": \"GPT-4.1\",\n",
" \"gpt_3_5_turbo\": \"GPT-3.5-turbo\",\n",
"}\n",
"\n",
"# -------------------------------------------------\n",
"# Load candidate totals per app\n",
"# CSV must contain columns:\n",
"# app_code,total_columns\n",
"# -------------------------------------------------\n",
"def load_candidate_totals(csv_path):\n",
" totals = {}\n",
" with open(csv_path, newline=\"\", encoding=\"utf-8\") as f:\n",
" reader = csv.DictReader(f)\n",
" for row in reader:\n",
" app = row.get(\"app_code\")\n",
" total = row.get(\"total_columns\")\n",
" if app and total:\n",
" totals[app] = int(total)\n",
" return totals\n",
"\n",
"\n",
"# -------------------------------------------------\n",
"# Load scanned columns per app\n",
"# From normalized_results/<model>/app_level/app_level.jsonl\n",
"# -------------------------------------------------\n",
"def load_scanned_cols(app_level_jsonl: Path):\n",
" \"\"\"\n",
" Reads app_level.jsonl (corpus/app-level format).\n",
"\n",
" Your format:\n",
" {\n",
" \"db_path\": \"selectedDBs\\\\A1\",\n",
" ...\n",
" \"Num_of_source_columns_unique\": 29\n",
" }\n",
"\n",
" Returns:\n",
" { \"A1\": scanned_columns }\n",
" \"\"\"\n",
"\n",
" scanned = {}\n",
"\n",
" with app_level_jsonl.open(\"r\", encoding=\"utf-8\") as f:\n",
" for line in f:\n",
" line = line.strip()\n",
" if not line:\n",
" continue\n",
"\n",
" rec = json.loads(line)\n",
"\n",
" # ---- Extract app from db_path ----\n",
" db_path = rec.get(\"db_path\", \"\")\n",
" if not db_path:\n",
" continue\n",
"\n",
" app = Path(db_path).name # \"A1\"\n",
"\n",
" # ---- Use UNIQUE count for efficiency ----\n",
" n = rec.get(\"Num_of_source_columns_unique\")\n",
" if n is None:\n",
" n = rec.get(\"Num_of_source_columns\", 0)\n",
"\n",
" try:\n",
" n = int(n)\n",
" except:\n",
" n = 0\n",
"\n",
" # Keep maximum seen per app\n",
" scanned[app] = max(scanned.get(app, 0), n)\n",
"\n",
" return scanned\n",
"\n",
"\n",
"\n",
"\n",
"# -------------------------------------------------\n",
"# Main computation\n",
"# -------------------------------------------------\n",
"candidate_totals = load_candidate_totals(CANDIDATE_CSV)\n",
"\n",
"results = []\n",
"\n",
"for slug, display_name in MODELS.items():\n",
"\n",
" app_jsonl = (\n",
" BASE_DIR\n",
" / \"normalized_results\"\n",
" / slug\n",
" / \"app_level\"\n",
" / \"app_level.jsonl\"\n",
" )\n",
"\n",
" if not app_jsonl.exists():\n",
" results.append((display_name, \"xx\", \"xx\"))\n",
" continue\n",
"\n",
" scanned_cols = load_scanned_cols(app_jsonl)\n",
"\n",
" examined_vals = []\n",
" reduction_vals = []\n",
"\n",
" for app, total_cols in candidate_totals.items():\n",
"\n",
" scanned = scanned_cols.get(app, 0)\n",
" examined_vals.append(scanned)\n",
"\n",
" if total_cols > 0:\n",
" reduction = 1 - (scanned / total_cols)\n",
" reduction_vals.append(reduction)\n",
"\n",
" if examined_vals and reduction_vals:\n",
" avg_examined = round(mean(examined_vals), 1)\n",
" avg_reduction = round(mean(reduction_vals) * 100, 1)\n",
" results.append((display_name, avg_examined, f\"{avg_reduction}\\\\%\"))\n",
" else:\n",
" results.append((display_name, \"xx\", \"xx\"))\n",
"\n",
"# -------------------------------------------------\n",
"# Print LaTeX table\n",
"# -------------------------------------------------\n",
"print(r\"\\begin{tabular}{|l|p{1.4cm}|p{1.8cm}|}\")\n",
"print(r\"\\hline\")\n",
"print(r\"\\textbf{Method/LLM} &\")\n",
"print(r\"\\textbf{Avg. Cols Examined} &\")\n",
"print(r\"\\textbf{Avg. Search Space Reduc.} \\\\\")\n",
"print(r\"\\hline\")\n",
"print(r\"bulk\\_extractor-v1.6 (baseline) & NA & 0.0\\% \\\\\")\n",
"print(r\"\\hline\")\n",
"\n",
"for name, cols, reduc in results:\n",
" print(f\"{name} & {cols} & {reduc} \\\\\\\\\")\n",
" print(r\"\\hline\")\n",
"\n",
"print(r\"\\end{tabular}\")\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "d17221d0",
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.14.0"
}
},
"nbformat": 4,
"nbformat_minor": 5
}

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

10
RQs/normalized_PII_results/ground_truth/db_level/PII_A1_commerce.jsonl
View File

@@ -1,5 +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": "EMAIL", "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": "PHONE", "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": "USERNAME", "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": "PERSON_NAME", "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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_A1_msgstore.jsonl
View File

File diff suppressed because one or more lines are too long

10
RQs/normalized_PII_results/ground_truth/db_level/PII_A1_wa.jsonl
View File

@@ -1,5 +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": "EMAIL", "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": "PHONE", "PII": ["9199037779", "8085096467", "5713349815", "6263678865", "6106046786", "7034241981", "5715917168", "2065937224", "5713298742", "8624338328", "8056377243", "2028177932", "2025692832", "9735203731", "7423794330"], "Num_of_PII": 15, "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": "USERNAME", "PII": ["clarke", "philips", "a_kalachikova", "elizabeth jones", "george", "morton", "latefire_eu", "mario.drapela", "dolgetta", "jacob emily", "thomas", "alethea", "peter", "imogen holman", "chloe rodriguez", "robert jane", "mason", "james matthew brown", "jalen", "official representative", "gilbert", "andie", "connor haggarty", "kop", "winona", "laurenncbrown", "gabriel", "sarah cox", "jesus delgado", "milicardis", "jackson", "dave", "benjamin", "elliot dowell", "mary garcia", "gaia", "renwbw", "jim anderson", "marit bonthuis", "rola nd", "marcel_juhas", "kevin destiny", "minardo gaspari", "stephanna smith", "ytliken", "thomas anderson", "pnod", "paco almeida", "oscar steven", "lockerbie do-vip", "admin", "roberto cadorin", "brandon addison", "matt galligan", "christellecamiller", "carlos", "marcinmis", "isabel abarca", "adam taylor", "antonio", "scrichpower", "hcibc", "gabi", "thomas arthur heber fearn", "augustin richard", "kamila", "aiden savannah", "edwardsaliendra", "jayden", "morgan", "harry", "kandicesledge", "jaroslaw machek", "juan pablo pesqueira", "mandy lauren🧙‍♀️", "gemmadorney", "zara", "marcos amorim", "amanda white", "mauro silvabarbosa", "felix davey", "aberudder", "isabella", "olivia", "felipe hernandez", "sophia", "miss you", "russell philby", "stephcoleman", "brian goudy", "liam thomas", "shawn hoxie", "carolasol_espinoza", "13135550002", "otto matik", "ruby", "million marketing", "jason steven", "‎shirley", "shehriyar ansari", "nole", "dungeon master", "alvin the alien", "perry", "angie", "bob the robot", "leo", "sally", "brian", "liv", "coco", "victor", "tamika", "becca", "lorena", "zach", "luiz", "bru", "thalia", "lily", "izzy", "max", "scarlett", "amber", "dylan", "jade", "billie", "jane austen", "meta", "leaura", "faxinezidohne", "patrickh34", "yauyauyauhen", "nathanmorris", "0jamesf", "italianmatters", "robm435", "loususi", "lanaire2023", "pet_the_bunny", "madmax_mgm", "reidback", "rennymorales", "yjr_fit.inba", "the_real_flockfam", "gibbogram", "almondeyezbitch", "visionz2turnt", "stars_hinemoon", "thisvillage_ijn", "patrick_c_doyle", "patrickvaxter", "humans_of_data", "airwicksol", "homan.jason", "nadhir_chiu_oficial", "brandonmcclainl", "geezdagawd", "superflysugar2024", "curia__", "psychicadvisor345", "lemieuxbrands", "james_macray_"], "Num_of_PII": 163, "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": "PERSON_NAME", "PII": ["russ philby", "don wayne", "whalen", "mary garcia", "joey", "hank", "goldie kahn", "abe rudder", "svetlana chernoff", "karen tate", "brian reynolds", "vladamir stravinsky", "otto", "toks", "bo", "voice mail", "capt don wayne", "ltc whalen", "sheri's vlog"], "Num_of_PII": 19, "source_columns": ["wa_address_book.display_name", "wa_address_book.given_name+family_name", "wa_address_book.sort_name", "wa_biz_profiles_linked_accounts_table.account_display_name", "wa_contacts.given_name+family_name", "wa_contacts.sort_name"], "Num_of_source_columns": 6}
{"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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_A2_core.jsonl
View File

@@ -1,5 +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": "EMAIL", "PII": ["sharononeil368@gmail.com"], "Num_of_PII": 1, "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": "PHONE", "PII": ["8624338329"], "Num_of_PII": 1, "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": "USERNAME", "PII": ["oneil3607"], "Num_of_PII": 1, "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": "PERSON_NAME", "PII": ["sharon oneil"], "Num_of_PII": 1, "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} {"db_path": "selectedDBs\\A2_core.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_A2_journal.jsonl
View File

@@ -1,5 +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": "EMAIL", "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": "PHONE", "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": "USERNAME", "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": "PERSON_NAME", "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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_A2_main.jsonl
View File

File diff suppressed because one or more lines are too long

10
RQs/normalized_PII_results/ground_truth/db_level/PII_A3_account1cache4.jsonl
View File

@@ -1,5 +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": "EMAIL", "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": "PHONE", "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": "USERNAME", "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": "PERSON_NAME", "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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_A3_account2cache4.jsonl
View File

@@ -1,5 +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": "EMAIL", "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": "PHONE", "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": "USERNAME", "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": "PERSON_NAME", "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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_A3_account3cache4.jsonl
View File

@@ -1,5 +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": "EMAIL", "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": "PHONE", "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": "USERNAME", "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": "PERSON_NAME", "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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_A4_gmm_myplaces.jsonl
View File

@@ -1,5 +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": "EMAIL", "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": "PHONE", "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": "USERNAME", "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": "PERSON_NAME", "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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_A4_gmm_storage.jsonl
View File

@@ -1,5 +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": "EMAIL", "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": "PHONE", "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": "USERNAME", "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": "PERSON_NAME", "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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14.jsonl
View File

@@ -1,5 +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": "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": "", "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": "PHONE", "PII": ["7423794330"], "Num_of_PII": 1, "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": "USERNAME", "PII": ["hmahalik"], "Num_of_PII": 1, "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": "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": "", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0} {"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}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_A5_SBrowser.jsonl
View File

@@ -1,5 +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": "EMAIL", "PII": ["sharononeil368@gmail.com"], "Num_of_PII": 1, "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": "PHONE", "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": "USERNAME", "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": "PERSON_NAME", "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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_A5_SBrowser2.jsonl
View File

@@ -1,5 +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": "EMAIL", "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": "PHONE", "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": "USERNAME", "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": "PERSON_NAME", "PII": ["sharon's s21"], "Num_of_PII": 1, "source_columns": ["bookmarks.device_name"], "Num_of_source_columns": 1}
{"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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_A5_searchengine.jsonl
View File

@@ -1,5 +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": "EMAIL", "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": "PHONE", "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": "USERNAME", "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": "PERSON_NAME", "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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_I1_CallHistory.jsonl
View File

@@ -1,5 +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": "EMAIL", "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": "PHONE", "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": "USERNAME", "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": "PERSON_NAME", "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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_I1_ChatStorage.jsonl
View File

@@ -1,5 +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": "EMAIL", "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": "PHONE", "PII": ["9716786701"], "Num_of_PII": 1, "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": "USERNAME", "PII": ["skinnyfatfarms", "emergencystream", "m.coinol.club", "anna_malina05"], "Num_of_PII": 4, "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": "PERSON_NAME", "PII": ["chad hunt", "toni yu", "charles finley", "ronen engler", "john raynolds", "jonathan reyes", "johnny good", "russell philby", "abe rudder", "finn", "john reynolds", "colin dacopps", "lisena gocaj", "andy sieg", "howell", "yaliweisi", "abner", "andy schweichert", "brian arseneau", "virginia", "christian justiniano", "jim wilson", "virginia benton", "lisena", "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", "brodie", "carlton", "ernesto torres cantú", "andy", "mr. andy", "anna malina", "rick", "otto", "reynolds", "sharon", "monica", "stephanie", "christian", "lisean", "mr. robechucks", "mr. robechucks raul", "mr robechucks", "thomas matthew crooks", "donald trump"], "Num_of_PII": 70, "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} {"db_path": "selectedDBs\\I1_ChatStorage.sqlite", "PII_type": "POSTAL_ADDRESS", "PII": ["8607 us-24, fort wayne, in 46804", "16 marina bay financial centre tower 2, 12 marina boulevard, singapore 018982", "12503 e via de palmas, chandler, az", "8500 peña blvd, denver, co"], "Num_of_PII": 4, "source_columns": ["zwamediaitem.ztitle", "zwamessage.ztext", "zwamessagedataitem.ztitle"], "Num_of_source_columns": 3}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_I1_ContactsV2.jsonl
View File

File diff suppressed because one or more lines are too long

10
RQs/normalized_PII_results/ground_truth/db_level/PII_I2_AddressBook.jsonl
View File

File diff suppressed because one or more lines are too long

10
RQs/normalized_PII_results/ground_truth/db_level/PII_I2_AddressBookImages.jsonl
View File

@@ -1,5 +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": "EMAIL", "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": "PHONE", "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": "USERNAME", "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": "PERSON_NAME", "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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_I3_sms.jsonl
View File

@@ -1,5 +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": "EMAIL", "PII": ["ottomatik1234@gmail.com", "aniulka616@hotmail.com", "josephphw3r5@outlook.com", "jreynoldscbdemo+this.is.secret@gmail.com", "jthskp131699@icloud.com", "sousiezorjrc@outlook.com", "sweeneyfrank594@gmail.com"], "Num_of_PII": 7, "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": "PHONE", "PII": ["9735203731", "2065937224", "2185715037", "2484345508", "3017157263", "4155346421", "4197574625", "4259796297", "4325353346", "4847353029", "5162879924", "7709195681", "7852533080", "7868720110", "8553965089", "8589330241", "8624338324", "8624338329", "9195796456", "9195796465", "9199037779", "1111288288", "1410100001", "2155817608", "5026417431", "8123654204", "9735203764", "9735203805", "9735203921"], "Num_of_PII": 29, "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": "USERNAME", "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": "PERSON_NAME", "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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_I4_CloudTabs.jsonl
View File

@@ -1,5 +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": "EMAIL", "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": "PHONE", "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": "USERNAME", "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": "PERSON_NAME", "PII": ["raven", "dickens", "dickins"], "Num_of_PII": 3, "source_columns": ["cloud_tabs.title"], "Num_of_source_columns": 1}
{"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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_I4_History.jsonl
View File

@@ -1,5 +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": "EMAIL", "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": "PHONE", "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": "USERNAME", "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": "PERSON_NAME", "PII": ["olivia newton-john"], "Num_of_PII": 1, "source_columns": ["history_tags.title"], "Num_of_source_columns": 1}
{"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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_I5_Calendar.jsonl
View File

@@ -1,5 +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": "EMAIL", "PII": ["ottomatik1234@gmail.com"], "Num_of_PII": 1, "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": "PHONE", "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": "USERNAME", "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": "PERSON_NAME", "PII": ["otto matik", "rosh hashanah", "ashura", "martin luther king jr. day", "otto matik’s birthday"], "Num_of_PII": 5, "source_columns": ["calendar.shared_owner_name", "calendaritem.summary", "identity.display_name", "store.owner_name"], "Num_of_source_columns": 4}
{"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} {"db_path": "selectedDBs\\I5_Calendar.sqlitedb", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

10
RQs/normalized_PII_results/ground_truth/db_level/PII_I5_Extras.jsonl
View File

@@ -1,5 +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": "EMAIL", "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": "PHONE", "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": "USERNAME", "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": "PERSON_NAME", "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": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Num_of_source_columns": 0}

BIN
RQs/usenixsecurity2026.pdf
View File

Binary file not shown.