Files
mobile-pii-discovery-agent/model_PII_results/gpt4o/PII_A2_journal_20260131T204142Z.jsonl

6 lines
8.8 KiB
JSON

{"db_path": "selectedDBs\\A2_journal.db", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Raw_rows_first_100": [], "Total_raw_rows": 0, "Exploration_sql": "SELECT path FROM journal WHERE path REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}' UNION ALL \nSELECT key FROM journal_entry WHERE key REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}' UNION ALL \nSELECT value_count FROM journal_entry WHERE value_count REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}' UNION ALL \nSELECT status FROM journal_entry WHERE status REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}' UNION ALL \nSELECT last_update_time FROM journal_entry WHERE last_update_time REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}' UNION ALL \nSELECT last_read_time FROM journal_entry WHERE last_read_time REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}' UNION ALL \nSELECT lock_count FROM journal_entry WHERE lock_count REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}' UNION ALL \nSELECT total_size FROM journal_entry WHERE total_size REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}' UNION ALL \nSELECT expiration FROM journal_entry WHERE expiration REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}' UNION ALL \nSELECT last_consumed_time FROM journal_entry WHERE last_consumed_time REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}' UNION ALL \nSELECT metadata FROM journal_entry WHERE metadata REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}';", "Extraction_sql": null, "PII_Prompt": "a unique identifier for a destination to which electronic mail (email) can be sent and received over the internet; examples include jane.doe@example.com, john.smith@provider.net, dev-team@startup.io, and user.name+label@domain.org"}
{"db_path": "selectedDBs\\A2_journal.db", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": [], "Raw_rows_first_100": [["9F3F465DC00D96696DDDFE0A946AAB99.khand_medium"], ["C730963C61386A34712C819CA25436C9.media"], ["70177660B739FDDF75DE848B97DC6A6E.edits"], ["03FD66A15523689AD035E1E2B1AD6DAE.chat_wallpaper_media"], ["D41F76126B39D1F7E7EC3D8FA4079D0F.discover_story_streaming_snap"], ["F05AD4876AFE7190FBF88E879238978A.discover_story_streaming_snap_ff"], ["67B685FF2948DC22416716E822D4F5A1.discover_story_streaming_snap_ff"], ["4EDFB389483E360F0CBA63F7A928FD39.discover_story_streaming_snap_ff"], ["629156B858FDF391C0639F2DE6933EEB.discover_story_streaming_snap_ff"], ["5A4F0362F54488AC2542C174F69C9A24.discover_story_streaming_snap_ff"]], "Total_raw_rows": 10, "Exploration_sql": "SELECT path FROM journal WHERE path REGEXP '\\+?[0-9]{1,4}[- .]?\\(?[0-9]{1,3}?\\)?[- .]?[0-9]{1,4}[- .]?[0-9]{1,4}[- .]?[0-9]{1,9}' UNION ALL SELECT key FROM journal_entry WHERE key REGEXP '\\+?[0-9]{1,4}[- .]?\\(?[0-9]{1,3}?\\)?[- .]?[0-9]{1,4}[- .]?[0-9]{1,4}[- .]?[0-9]{1,9}' UNION ALL SELECT value_count FROM journal_entry WHERE value_count REGEXP '\\+?[0-9]{1,4}[- .]?\\(?[0-9]{1,3}?\\)?[- .]?[0-9]{1,4}[- .]?[0-9]{1,4}[- .]?[0-9]{1,9}'", "Extraction_sql": null, "PII_Prompt": "a US phone number is a 10-digit NANP number (area code + exchange + line) that may be written as 2023133725, 202-313-3725, (202) 313-3725, 202.313.3725, +1 202 313 3725, or 1-202-313-3725"}
{"db_path": "selectedDBs\\A2_journal.db", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Raw_rows_first_100": [["/data/data/com.snapchat.android/files/file_manager/BLOOPS_STICKER"], ["/data/data/com.snapchat.android/files/file_manager/Bitmoji_Preview"], ["/data/data/com.snapchat.android/files/file_manager/LENS_ASSET_CONTENT_TYPE_NAME"], ["/data/data/com.snapchat.android/files/file_manager/Live_Mirror_Model"], ["/data/data/com.snapchat.android/files/file_manager/Login_Kit_Privacy"], ["/data/data/com.snapchat.android/files/file_manager/MUSIC_GENERIC_ASSET_TYPE"], ["/data/data/com.snapchat.android/files/file_manager/Maps_Kashmir"], ["/data/data/com.snapchat.android/files/file_manager/Maps_WorldEffects"], ["/data/data/com.snapchat.android/files/file_manager/Perception"], ["/data/data/com.snapchat.android/files/file_manager/PerceptionMl"], ["12DB3FD3B46FC8F9DD60F79CB359FBFE.khand_medium"], ["9F3F465DC00D96696DDDFE0A946AAB99.khand_medium"], ["2FC6ABAAFF969A947FAB4E52FE0971FC.thumbnail"], ["C730963C61386A34712C819CA25436C9.media"], ["70177660B739FDDF75DE848B97DC6A6E.edits"], ["03FD66A15523689AD035E1E2B1AD6DAE.chat_wallpaper_media"], ["D41F76126B39D1F7E7EC3D8FA4079D0F.discover_story_streaming_snap"], ["5BBE52CE6D0010CB50CA3221C4741E7D.discover_story_streaming_snap_ff"], ["F05AD4876AFE7190FBF88E879238978A.discover_story_streaming_snap_ff"], ["67B685FF2948DC22416716E822D4F5A1.discover_story_streaming_snap_ff"]], "Total_raw_rows": 20, "Exploration_sql": "SELECT path FROM journal WHERE path REGEXP '\\b[a-zA-Z][a-zA-Z0-9._-]{2,51}\\b' UNION ALL SELECT key FROM journal_entry WHERE key REGEXP '\\b[a-zA-Z][a-zA-Z0-9._-]{2,51}\\b'", "Extraction_sql": null, "PII_Prompt": "a username is a short textual identifier chosen by a user to represent their account or public handle within an application or service it is stored as plain text contains no whitespace does not include a domain component and is intended for human recognition rather than internal system uniqueness"}
{"db_path": "selectedDBs\\A2_journal.db", "PII_type": "PERSON_NAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Raw_rows_first_100": [["/data/data/com.snapchat.android/files/file_manager/BLOOPS_STICKER"], ["/data/data/com.snapchat.android/files/file_manager/Bitmoji_Preview"], ["/data/data/com.snapchat.android/files/file_manager/LENS_ASSET_CONTENT_TYPE_NAME"], ["/data/data/com.snapchat.android/files/file_manager/Live_Mirror_Model"], ["/data/data/com.snapchat.android/files/file_manager/Login_Kit_Privacy"], ["/data/data/com.snapchat.android/files/file_manager/MUSIC_GENERIC_ASSET_TYPE"], ["/data/data/com.snapchat.android/files/file_manager/Maps_Kashmir"], ["/data/data/com.snapchat.android/files/file_manager/Maps_WorldEffects"], ["/data/data/com.snapchat.android/files/file_manager/Perception"], ["/data/data/com.snapchat.android/files/file_manager/PerceptionMl"], ["12DB3FD3B46FC8F9DD60F79CB359FBFE.khand_medium"], ["9F3F465DC00D96696DDDFE0A946AAB99.khand_medium"], ["2FC6ABAAFF969A947FAB4E52FE0971FC.thumbnail"], ["C730963C61386A34712C819CA25436C9.media"], ["70177660B739FDDF75DE848B97DC6A6E.edits"], ["03FD66A15523689AD035E1E2B1AD6DAE.chat_wallpaper_media"], ["D41F76126B39D1F7E7EC3D8FA4079D0F.discover_story_streaming_snap"], ["5BBE52CE6D0010CB50CA3221C4741E7D.discover_story_streaming_snap_ff"], ["F05AD4876AFE7190FBF88E879238978A.discover_story_streaming_snap_ff"], ["67B685FF2948DC22416716E822D4F5A1.discover_story_streaming_snap_ff"]], "Total_raw_rows": 20, "Exploration_sql": "SELECT path FROM journal WHERE path REGEXP '[A-Za-z][A-Za-z\\s\\.\\-]{1,50}' \nUNION ALL \nSELECT key FROM journal_entry WHERE key REGEXP '[A-Za-z][A-Za-z\\s\\.\\-]{1,50}' \nUNION ALL \nSELECT value_count FROM journal_entry WHERE value_count REGEXP '[A-Za-z][A-Za-z\\s\\.\\-]{1,50}' \nUNION ALL \nSELECT status FROM journal_entry WHERE status REGEXP '[A-Za-z][A-Za-z\\s\\.\\-]{1,50}' \nUNION ALL \nSELECT metadata FROM journal_entry WHERE metadata REGEXP '[A-Za-z][A-Za-z\\s\\.\\-]{1,50}';", "Extraction_sql": null, "PII_Prompt": "a loosely structured human name-like strings that typically consist of a first name, a first name and a last name, and may also include middle names, initials, prefixes (e.g., Mr., Dr.), and suffixes (e.g., Jr., Sr.)"}
{"db_path": "selectedDBs\\A2_journal.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Raw_rows_first_100": [["/data/data/com.snapchat.android/files/file_manager/map-place-icons"], ["/data/data/com.snapchat.android/files/file_manager/place-profile-annotation"], ["/data/data/com.snapchat.android/files/file_manager/place-profile-info"]], "Total_raw_rows": 3, "Exploration_sql": "SELECT path FROM journal WHERE path REGEXP '(?i)\\b(?:p\\.?\\s*o\\.?\\s*box|post\\s+office\\s+box|ave\\.?|avenue|st\\.?|street|rd\\.?|road|blvd\\.?|boulevard|dr\\.?|drive|ln\\.?|lane|ct\\.?|court|pl\\.?|place|way|pkwy\\.?|parkway|cir\\.?|circle|ter\\.?|terrace|hwy\\.?|highway|trl\\.?|trail|sq\\.?|square|pike|loop|run|walk|path|byp\\.?|bypass|(?:n|s|e|w|ne|nw|se|sw)\\b)\\b' \nUNION ALL \nSELECT key FROM journal_entry WHERE key REGEXP '(?i)\\b(?:p\\.?\\s*o\\.?\\s*box|post\\s+office\\s+box|ave\\.?|avenue|st\\.?|street|rd\\.?|road|blvd\\.?|boulevard|dr\\.?|drive|ln\\.?|lane|ct\\.?|court|pl\\.?|place|way|pkwy\\.?|parkway|cir\\.?|circle|ter\\.?|terrace|hwy\\.?|highway|trl\\.?|trail|sq\\.?|square|pike|loop|run|walk|path|byp\\.?|bypass|(?:n|s|e|w|ne|nw|se|sw)\\b)\\b'", "Extraction_sql": null, "PII_Prompt": "a US postal address is a street-level mailing location in the United States, commonly appearing as a street name and suffix (e.g., 'Market St') optionally with a street number (e.g., '1500 Market St'), unit, city/state, ZIP, or a PO Box (e.g., 'P.O. Box 123')"}