Browse Prior Art Database

Method and System for Detecting Abusive Users in Community Question Answering Platforms

IP.com Disclosure Number: IPCOM000240232D
Publication Date: 2015-Jan-14

Publishing Venue

The IP.com Prior Art Database

Related People

Nicolas Kourtellis: INVENTOR [+3]

Abstract

A method and system is disclosed for detecting abusive users in community question answering platforms. The method and system automatically differentiates between abusive and deviant users to increase user retention and engagement. Thus, the method and system enables increase in higher quality content in the community question answering platforms.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 16% of the total text.

Method and System for Detecting Abusive Users in Community Question Answering Platforms

Abstract

A method and system is disclosed for detecting abusive users in community question answering platforms.  The method and system automatically differentiates between abusive and deviant users to increase user retention and engagement.  Thus, the method and system enables increase in higher quality content in the community question answering platforms.

Description

For any online platform that relies on user generated content, there is a need to have content of high quality.  Unfortunately, at times, information on such platforms tends to be of low quality such as offensive or abusive information.  Generally, abusive users can greatly affect content quality.  In a mobile device with reduced display space and core functionalities, abusive content leads to negative impact on the user experience.  The negative impact can frustrate users who end up perceiving a platform as a waste of time, bandwidth and battery.  However, deviants but not abusive users are those who take risks by engaging in controversial topics and report low quality content without fear of retaliation.  Such deviants can increase engagement from other users and thus drive more traffic to the platform.

Disclosed is a method and system for detecting abusive users in community question answering platforms.  The method and system automatically differentiates between abusive and deviant users to increase user retention and engagement.  Thus, the method and system enables increase in higher quality content in the community question answering platforms.  The method and system analyzes activity of users who post content that violates community guidelines of a content publisher and automatically ranks the analyzed users on a deviance scale.  Depending on the severity of the violation, the method and system can propose to the platform’s administrator suspension of particular user accounts.

The method and system receives input, which is a set of users with ego networks, activity logs and the reports filed against the set of users for abusive content.  Based on the input, the method and system classifies each user as "fair", "deviant" and "should be suspended".  The classification is carried out based on a score, wherein the score reflects the deviation of the user from community norms.

The method and system analyzes activities of a user in the platform.  The activities can include the number of questions and answers posted by the user, and the number of abuse reports that the user received for that content posted.  After analyzing, the method and system computes a deviance score for that user with the following formula:

DevianceACT (u) =YACT(u)-YACT(u)

wherein, YACT(u) is the number of correct abuse reports received for user u for the activity (questions/answers) of the user and YACT(u) is the expected number of correct abuse reports to be received by user u for that amount of ac...