Data description and request procedure for apps that can be used for intimate partner surveillance

IPVSpyware.md

The Spyware Used in Intimate Partner Violence

If you want to get the list of apps we found that can be used for intimate partner surveillance (IPS), please contact at (rahul@cs.cornell.edu) with your affiliation and a brief description of your purpose for obtaining the data.

app-info.db

The app-info.db contains all the app information we have crawled from Play Store and App Store, and output of our ML and manual classification. This is a sqlite database you can use sqlite3 in terminal or Sqlite Viewer type browser extension to view the datbase.

The list of apps relevant for IPS is listed in appflags. The column store are {playstore, appstore, offstore}. AppId column (appId) is an unique id assigned by the application stores. The flag denotes whether or not the app is a overt spyware (that is they advertise for spying on someone), or dual-use (apps that are not built for spying, but can be very easily repurposed for spying on someone, especially an intimate partner).

More information about apps are provided in apps table, where
most interesting collumns are appId, summary, title, description, relevant, ml_score, and class.

• ml_score is the score assigned by our ML classifier for flagging apps as relevant for IPS. relevant column is post processing of ML data by human analysts. The final list of flagged apps are in appflags, and I would recommend using that table to determine whether or not an app is relevant.
• class column sometimes contain a manually given class of the app based on their description, for example, parental control, child tracking, ff tracker (friends and family tracker), etc. The flags given in this column is not through a very thorough methodology. It's an intial attempt to classify all the apps we have seen. However, I am all ears for suggestion to improve it. will be happy to answer question if you have about this column. The column

sqlite> .schema apps
CREATE TABLE apps (
	"appId" TEXT, 
	application_icon TEXT, 
	appwebsite TEXT, 
	class TEXT, 
	comment TEXT, 
	contentrating TEXT, 
	description TEXT, 
	developerwebsite TEXT, 
	filename TEXT, 
	genreid TEXT, 
	genres TEXT, 
	id FLOAT, 
	launchable_activity_icon TEXT, 
	launchable_activity_label TEXT, 
	launchable_activity_name TEXT, 
	ml_score FLOAT, 
	package_platformbuildversionname TEXT, 
	package_versioncode FLOAT, 
	package_versionname TEXT, 
	permissions TEXT, 
	price FLOAT, 
	relevant TEXT, 
	relevant_old TEXT, 
	reviewer1_comment TEXT, 
	sha256sum TEXT, 
	store TEXT, 
	subclass TEXT, 
	summary TEXT, 
	title TEXT, 
	y FLOAT
);

sqlite> .schema appflags
CREATE TABLE appflags(
  "appId" TEXT,
  "store" TEXT,
  "flag" TEXT,
  "title" TEXT
);

We have built a basic scaning tool to scan phones with these list of flagged apps.

You can get more informaiton about our research group https://www.ipvtechresearch.org/.