Browse Prior Art Database

Cognitive comment analysis for preventing legal actions

IP.com Disclosure Number: IPCOM000246853D
Publication Date: 2016-Jul-07
Document File: 1 page(s) / 23K

Publishing Venue

The IP.com Prior Art Database

Abstract

A method for cognitive comment analysis in order to prevent legal actions

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 86% of the total text.

Page 01 of 1

Cognitive comment analysis for preventing legal actions

Disclosed is a method for cognitive comment analysis in order to prevent legal actions.

Users often are asked to leave their comments/opinions on auction sites/e-commerce shops and similar. Frequently, negative comments, when acting in emotions may be cause of legal actions against commenting users. Our solution prevents/warns users that their comment may for example be offensive or lead to legal actions otherwise. Comment text is passed to "cognitive classifier" that calculates how likely it is that it may be inappropriate, and then if likelihood is high

warns user that posting such comment *may* lead to legal actions. System is positioned between business logic and UI but may as well be part of business logic. System would work in two phases (flow shown on figure 1.):


1. Learning

Already known comments that lead to legal actions and containing offensive words

would be shown to the system. This may be continuous or periodic process to keep the system "knowledge" fresh and accurate (it may be triggerd if new comment (that

was undetected or detected and posted anyway) lead to legal actions).


2. Classification


System would be feed with new comments issued by users and would reply with results, UI layer would display a warning.

Fig. 1


Embodiment:
Someone buys something on auction site. And is not happy with what she/he received (item differs from auction description). Thus in high emotions tries to post comment...