From 8e6158bfd7bb3a7dc8625b454d5a6f6f7907f644 Mon Sep 17 00:00:00 2001 From: Frank Xu Date: Wed, 28 Jan 2026 20:16:13 -0500 Subject: [PATCH] change name to match models --- ....ipynb => RQ0_results_normalization.ipynb} | 31 +++++++++++++++++-- .../PII_A1_commerce_20260127T175911Z.jsonl | 0 .../PII_A1_msgstore_20260127T180043Z.jsonl | 0 .../PII_A1_wa_20260127T180213Z.jsonl | 0 .../PII_A2_core_20260127T180339Z.jsonl | 0 .../PII_A2_journal_20260127T180440Z.jsonl | 0 .../PII_A2_main_20260127T180710Z.jsonl | 0 ...I_A3_account1cache4_20260127T180745Z.jsonl | 0 ...I_A3_account2cache4_20260127T180821Z.jsonl | 0 ...I_A3_account3cache4_20260127T180857Z.jsonl | 0 ...PII_A4_gmm_myplaces_20260127T180935Z.jsonl | 0 .../PII_A4_gmm_storage_20260127T181014Z.jsonl | 0 ...l.com_com.google_14_20260127T181121Z.jsonl | 0 .../PII_A5_SBrowser2_20260127T181345Z.jsonl | 0 .../PII_A5_SBrowser_20260127T181239Z.jsonl | 0 ...PII_A5_searchengine_20260127T181446Z.jsonl | 0 .../PII_I1_CallHistory_20260127T181557Z.jsonl | 0 .../PII_I1_ChatStorage_20260127T181731Z.jsonl | 0 .../PII_I1_ContactsV2_20260127T182906Z.jsonl | 0 ...2_AddressBookImages_20260127T183526Z.jsonl | 0 .../PII_I2_AddressBook_20260127T183457Z.jsonl | 0 .../PII_I3_sms_20260127T183606Z.jsonl | 0 .../PII_I4_CloudTabs_20260127T183643Z.jsonl | 0 .../PII_I4_History_20260127T183727Z.jsonl | 0 .../PII_I5_Calendar_20260127T183815Z.jsonl | 0 .../PII_I5_Extras_20260127T183857Z.jsonl | 0 .../PII_A1_commerce.jsonl | 2 +- .../PII_A1_msgstore.jsonl | 2 +- RQs/ground_truth_normalized/PII_A1_wa.jsonl | 16 +++++----- RQs/ground_truth_normalized/PII_A2_core.jsonl | 8 ++--- .../PII_A2_journal.jsonl | 2 +- RQs/ground_truth_normalized/PII_A2_main.jsonl | 20 ++++++------ .../PII_A3_account1cache4.jsonl | 2 +- .../PII_A3_account2cache4.jsonl | 2 +- .../PII_A3_account3cache4.jsonl | 2 +- .../PII_A4_gmm_myplaces.jsonl | 2 +- .../PII_A4_gmm_storage.jsonl | 2 +- ...arononeil368@gmail.com_com.google_14.jsonl | 12 +++---- .../PII_A5_SBrowser.jsonl | 6 ++-- .../PII_A5_SBrowser2.jsonl | 2 +- .../PII_A5_searchengine.jsonl | 2 +- .../PII_I1_CallHistory.jsonl | 2 +- .../PII_I1_ChatStorage.jsonl | 8 ++--- .../PII_I1_ContactsV2.jsonl | 4 +-- .../PII_I2_AddressBook.jsonl | 12 ++++--- .../PII_I2_AddressBookImages.jsonl | 2 +- RQs/ground_truth_normalized/PII_I3_sms.jsonl | 12 +++---- .../PII_I4_CloudTabs.jsonl | 2 +- .../PII_I4_History.jsonl | 2 +- .../PII_I5_Calendar.jsonl | 11 ++++--- .../PII_I5_Extras.jsonl | 2 +- 51 files changed, 100 insertions(+), 70 deletions(-) rename RQs/RQ0/{RQ0_batch_results_normalization.ipynb => RQ0_results_normalization.ipynb} (72%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_A1_commerce_20260127T175911Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_A1_msgstore_20260127T180043Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_A1_wa_20260127T180213Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_A2_core_20260127T180339Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_A2_journal_20260127T180440Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_A2_main_20260127T180710Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_A3_account1cache4_20260127T180745Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_A3_account2cache4_20260127T180821Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_A3_account3cache4_20260127T180857Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_A4_gmm_myplaces_20260127T180935Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_A4_gmm_storage_20260127T181014Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14_20260127T181121Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_A5_SBrowser2_20260127T181345Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_A5_SBrowser_20260127T181239Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_A5_searchengine_20260127T181446Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_I1_CallHistory_20260127T181557Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_I1_ChatStorage_20260127T181731Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_I1_ContactsV2_20260127T182906Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_I2_AddressBookImages_20260127T183526Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_I2_AddressBook_20260127T183457Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_I3_sms_20260127T183606Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_I4_CloudTabs_20260127T183643Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_I4_History_20260127T183727Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_I5_Calendar_20260127T183815Z.jsonl (100%) rename RQs/{batch_results_normalized => batch_results_gpt4o_normalized}/PII_I5_Extras_20260127T183857Z.jsonl (100%) diff --git a/RQs/RQ0/RQ0_batch_results_normalization.ipynb b/RQs/RQ0/RQ0_results_normalization.ipynb similarity index 72% rename from RQs/RQ0/RQ0_batch_results_normalization.ipynb rename to RQs/RQ0/RQ0_results_normalization.ipynb index d61b665..da475b2 100644 --- a/RQs/RQ0/RQ0_batch_results_normalization.ipynb +++ b/RQs/RQ0/RQ0_results_normalization.ipynb @@ -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", diff --git a/RQs/batch_results_normalized/PII_A1_commerce_20260127T175911Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_A1_commerce_20260127T175911Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_A1_commerce_20260127T175911Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_A1_commerce_20260127T175911Z.jsonl diff --git a/RQs/batch_results_normalized/PII_A1_msgstore_20260127T180043Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_A1_msgstore_20260127T180043Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_A1_msgstore_20260127T180043Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_A1_msgstore_20260127T180043Z.jsonl diff --git a/RQs/batch_results_normalized/PII_A1_wa_20260127T180213Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_A1_wa_20260127T180213Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_A1_wa_20260127T180213Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_A1_wa_20260127T180213Z.jsonl diff --git a/RQs/batch_results_normalized/PII_A2_core_20260127T180339Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_A2_core_20260127T180339Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_A2_core_20260127T180339Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_A2_core_20260127T180339Z.jsonl diff --git a/RQs/batch_results_normalized/PII_A2_journal_20260127T180440Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_A2_journal_20260127T180440Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_A2_journal_20260127T180440Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_A2_journal_20260127T180440Z.jsonl diff --git a/RQs/batch_results_normalized/PII_A2_main_20260127T180710Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_A2_main_20260127T180710Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_A2_main_20260127T180710Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_A2_main_20260127T180710Z.jsonl diff --git a/RQs/batch_results_normalized/PII_A3_account1cache4_20260127T180745Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_A3_account1cache4_20260127T180745Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_A3_account1cache4_20260127T180745Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_A3_account1cache4_20260127T180745Z.jsonl diff --git a/RQs/batch_results_normalized/PII_A3_account2cache4_20260127T180821Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_A3_account2cache4_20260127T180821Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_A3_account2cache4_20260127T180821Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_A3_account2cache4_20260127T180821Z.jsonl diff --git a/RQs/batch_results_normalized/PII_A3_account3cache4_20260127T180857Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_A3_account3cache4_20260127T180857Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_A3_account3cache4_20260127T180857Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_A3_account3cache4_20260127T180857Z.jsonl diff --git a/RQs/batch_results_normalized/PII_A4_gmm_myplaces_20260127T180935Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_A4_gmm_myplaces_20260127T180935Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_A4_gmm_myplaces_20260127T180935Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_A4_gmm_myplaces_20260127T180935Z.jsonl diff --git a/RQs/batch_results_normalized/PII_A4_gmm_storage_20260127T181014Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_A4_gmm_storage_20260127T181014Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_A4_gmm_storage_20260127T181014Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_A4_gmm_storage_20260127T181014Z.jsonl diff --git a/RQs/batch_results_normalized/PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14_20260127T181121Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14_20260127T181121Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14_20260127T181121Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14_20260127T181121Z.jsonl diff --git a/RQs/batch_results_normalized/PII_A5_SBrowser2_20260127T181345Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_A5_SBrowser2_20260127T181345Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_A5_SBrowser2_20260127T181345Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_A5_SBrowser2_20260127T181345Z.jsonl diff --git a/RQs/batch_results_normalized/PII_A5_SBrowser_20260127T181239Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_A5_SBrowser_20260127T181239Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_A5_SBrowser_20260127T181239Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_A5_SBrowser_20260127T181239Z.jsonl diff --git a/RQs/batch_results_normalized/PII_A5_searchengine_20260127T181446Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_A5_searchengine_20260127T181446Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_A5_searchengine_20260127T181446Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_A5_searchengine_20260127T181446Z.jsonl diff --git a/RQs/batch_results_normalized/PII_I1_CallHistory_20260127T181557Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_I1_CallHistory_20260127T181557Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_I1_CallHistory_20260127T181557Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_I1_CallHistory_20260127T181557Z.jsonl diff --git a/RQs/batch_results_normalized/PII_I1_ChatStorage_20260127T181731Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_I1_ChatStorage_20260127T181731Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_I1_ChatStorage_20260127T181731Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_I1_ChatStorage_20260127T181731Z.jsonl diff --git a/RQs/batch_results_normalized/PII_I1_ContactsV2_20260127T182906Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_I1_ContactsV2_20260127T182906Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_I1_ContactsV2_20260127T182906Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_I1_ContactsV2_20260127T182906Z.jsonl diff --git a/RQs/batch_results_normalized/PII_I2_AddressBookImages_20260127T183526Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_I2_AddressBookImages_20260127T183526Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_I2_AddressBookImages_20260127T183526Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_I2_AddressBookImages_20260127T183526Z.jsonl diff --git a/RQs/batch_results_normalized/PII_I2_AddressBook_20260127T183457Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_I2_AddressBook_20260127T183457Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_I2_AddressBook_20260127T183457Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_I2_AddressBook_20260127T183457Z.jsonl diff --git a/RQs/batch_results_normalized/PII_I3_sms_20260127T183606Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_I3_sms_20260127T183606Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_I3_sms_20260127T183606Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_I3_sms_20260127T183606Z.jsonl diff --git a/RQs/batch_results_normalized/PII_I4_CloudTabs_20260127T183643Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_I4_CloudTabs_20260127T183643Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_I4_CloudTabs_20260127T183643Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_I4_CloudTabs_20260127T183643Z.jsonl diff --git a/RQs/batch_results_normalized/PII_I4_History_20260127T183727Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_I4_History_20260127T183727Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_I4_History_20260127T183727Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_I4_History_20260127T183727Z.jsonl diff --git a/RQs/batch_results_normalized/PII_I5_Calendar_20260127T183815Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_I5_Calendar_20260127T183815Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_I5_Calendar_20260127T183815Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_I5_Calendar_20260127T183815Z.jsonl diff --git a/RQs/batch_results_normalized/PII_I5_Extras_20260127T183857Z.jsonl b/RQs/batch_results_gpt4o_normalized/PII_I5_Extras_20260127T183857Z.jsonl similarity index 100% rename from RQs/batch_results_normalized/PII_I5_Extras_20260127T183857Z.jsonl rename to RQs/batch_results_gpt4o_normalized/PII_I5_Extras_20260127T183857Z.jsonl diff --git a/RQs/ground_truth_normalized/PII_A1_commerce.jsonl b/RQs/ground_truth_normalized/PII_A1_commerce.jsonl index ef4928c..ee0cf5a 100644 --- a/RQs/ground_truth_normalized/PII_A1_commerce.jsonl +++ b/RQs/ground_truth_normalized/PII_A1_commerce.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_A1_msgstore.jsonl b/RQs/ground_truth_normalized/PII_A1_msgstore.jsonl index d6c1c8a..14b7eb0 100644 --- a/RQs/ground_truth_normalized/PII_A1_msgstore.jsonl +++ b/RQs/ground_truth_normalized/PII_A1_msgstore.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_A1_wa.jsonl b/RQs/ground_truth_normalized/PII_A1_wa.jsonl index e4850f1..0297fe6 100644 --- a/RQs/ground_truth_normalized/PII_A1_wa.jsonl +++ b/RQs/ground_truth_normalized/PII_A1_wa.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_A2_core.jsonl b/RQs/ground_truth_normalized/PII_A2_core.jsonl index c2585ea..708055c 100644 --- a/RQs/ground_truth_normalized/PII_A2_core.jsonl +++ b/RQs/ground_truth_normalized/PII_A2_core.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_A2_journal.jsonl b/RQs/ground_truth_normalized/PII_A2_journal.jsonl index f675743..781f982 100644 --- a/RQs/ground_truth_normalized/PII_A2_journal.jsonl +++ b/RQs/ground_truth_normalized/PII_A2_journal.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_A2_main.jsonl b/RQs/ground_truth_normalized/PII_A2_main.jsonl index 198e79f..373ce0d 100644 --- a/RQs/ground_truth_normalized/PII_A2_main.jsonl +++ b/RQs/ground_truth_normalized/PII_A2_main.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_A3_account1cache4.jsonl b/RQs/ground_truth_normalized/PII_A3_account1cache4.jsonl index 14b20f2..df8f3dd 100644 --- a/RQs/ground_truth_normalized/PII_A3_account1cache4.jsonl +++ b/RQs/ground_truth_normalized/PII_A3_account1cache4.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_A3_account2cache4.jsonl b/RQs/ground_truth_normalized/PII_A3_account2cache4.jsonl index c1dc63c..1bb212f 100644 --- a/RQs/ground_truth_normalized/PII_A3_account2cache4.jsonl +++ b/RQs/ground_truth_normalized/PII_A3_account2cache4.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_A3_account3cache4.jsonl b/RQs/ground_truth_normalized/PII_A3_account3cache4.jsonl index b802427..67d5bb9 100644 --- a/RQs/ground_truth_normalized/PII_A3_account3cache4.jsonl +++ b/RQs/ground_truth_normalized/PII_A3_account3cache4.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_A4_gmm_myplaces.jsonl b/RQs/ground_truth_normalized/PII_A4_gmm_myplaces.jsonl index 82b1e89..f490fcf 100644 --- a/RQs/ground_truth_normalized/PII_A4_gmm_myplaces.jsonl +++ b/RQs/ground_truth_normalized/PII_A4_gmm_myplaces.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_A4_gmm_storage.jsonl b/RQs/ground_truth_normalized/PII_A4_gmm_storage.jsonl index ce57d30..032e5af 100644 --- a/RQs/ground_truth_normalized/PII_A4_gmm_storage.jsonl +++ b/RQs/ground_truth_normalized/PII_A4_gmm_storage.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14.jsonl b/RQs/ground_truth_normalized/PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14.jsonl index d78831c..23b21a4 100644 --- a/RQs/ground_truth_normalized/PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14.jsonl +++ b/RQs/ground_truth_normalized/PII_A4_peopleCache_sharononeil368@gmail.com_com.google_14.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_A5_SBrowser.jsonl b/RQs/ground_truth_normalized/PII_A5_SBrowser.jsonl index 6bcf6f4..cd32936 100644 --- a/RQs/ground_truth_normalized/PII_A5_SBrowser.jsonl +++ b/RQs/ground_truth_normalized/PII_A5_SBrowser.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_A5_SBrowser2.jsonl b/RQs/ground_truth_normalized/PII_A5_SBrowser2.jsonl index b84ba31..dbeec21 100644 --- a/RQs/ground_truth_normalized/PII_A5_SBrowser2.jsonl +++ b/RQs/ground_truth_normalized/PII_A5_SBrowser2.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_A5_searchengine.jsonl b/RQs/ground_truth_normalized/PII_A5_searchengine.jsonl index 3d5b8a4..27db3c8 100644 --- a/RQs/ground_truth_normalized/PII_A5_searchengine.jsonl +++ b/RQs/ground_truth_normalized/PII_A5_searchengine.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_I1_CallHistory.jsonl b/RQs/ground_truth_normalized/PII_I1_CallHistory.jsonl index 188cf7b..2699cac 100644 --- a/RQs/ground_truth_normalized/PII_I1_CallHistory.jsonl +++ b/RQs/ground_truth_normalized/PII_I1_CallHistory.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_I1_ChatStorage.jsonl b/RQs/ground_truth_normalized/PII_I1_ChatStorage.jsonl index 01f31ca..bd8167b 100644 --- a/RQs/ground_truth_normalized/PII_I1_ChatStorage.jsonl +++ b/RQs/ground_truth_normalized/PII_I1_ChatStorage.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_I1_ContactsV2.jsonl b/RQs/ground_truth_normalized/PII_I1_ContactsV2.jsonl index 27ba427..49818bd 100644 --- a/RQs/ground_truth_normalized/PII_I1_ContactsV2.jsonl +++ b/RQs/ground_truth_normalized/PII_I1_ContactsV2.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_I2_AddressBook.jsonl b/RQs/ground_truth_normalized/PII_I2_AddressBook.jsonl index c93a193..af7bda3 100644 --- a/RQs/ground_truth_normalized/PII_I2_AddressBook.jsonl +++ b/RQs/ground_truth_normalized/PII_I2_AddressBook.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_I2_AddressBookImages.jsonl b/RQs/ground_truth_normalized/PII_I2_AddressBookImages.jsonl index ea15947..c24a700 100644 --- a/RQs/ground_truth_normalized/PII_I2_AddressBookImages.jsonl +++ b/RQs/ground_truth_normalized/PII_I2_AddressBookImages.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_I3_sms.jsonl b/RQs/ground_truth_normalized/PII_I3_sms.jsonl index 6e3efec..2c6ec0a 100644 --- a/RQs/ground_truth_normalized/PII_I3_sms.jsonl +++ b/RQs/ground_truth_normalized/PII_I3_sms.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_I4_CloudTabs.jsonl b/RQs/ground_truth_normalized/PII_I4_CloudTabs.jsonl index d9f7088..e40af70 100644 --- a/RQs/ground_truth_normalized/PII_I4_CloudTabs.jsonl +++ b/RQs/ground_truth_normalized/PII_I4_CloudTabs.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_I4_History.jsonl b/RQs/ground_truth_normalized/PII_I4_History.jsonl index c8167a3..83367c4 100644 --- a/RQs/ground_truth_normalized/PII_I4_History.jsonl +++ b/RQs/ground_truth_normalized/PII_I4_History.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_I5_Calendar.jsonl b/RQs/ground_truth_normalized/PII_I5_Calendar.jsonl index 78c9198..053119e 100644 --- a/RQs/ground_truth_normalized/PII_I5_Calendar.jsonl +++ b/RQs/ground_truth_normalized/PII_I5_Calendar.jsonl @@ -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} diff --git a/RQs/ground_truth_normalized/PII_I5_Extras.jsonl b/RQs/ground_truth_normalized/PII_I5_Extras.jsonl index d64dd2f..bd01486 100644 --- a/RQs/ground_truth_normalized/PII_I5_Extras.jsonl +++ b/RQs/ground_truth_normalized/PII_I5_Extras.jsonl @@ -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}