Identifying Sentiment-based Contradictions

Abstract

Contradiction Analysis is a relatively new multidisciplinary and complex area with the main goal of identifying contradictory pieces of text.
It can be addressed from the perspectives of different research areas such as Natural Language Processing, Opinion Mining, Information Retrieval, and Information Extraction.
This article focuses on the problem of detecting sentiment-based contradictions which occur in the sentences of a given review text.
Unlike other types of contradictions, the detection of sentiment-based contradictions can be tackled as a post-processing step in the traditional sentiment analysis task.
In this context, we adapted and extended an existing contradiction analysis framework by filtering its results to remove the reviews that are erroneously labeled as contradictory.
The filtering method is based on two simple term similarity algorithms which relies on sets of known positive and negative words.
An experimental evaluation on real product reviews has shown proportional improvements of up to 30% in classification accuracy and 26% in the precision of contradiction detection by considering an manual selection of positive and negative words. When the sets of positive and negative words are automatically selected, the improvements are of up to 22% in classification accuracy and 37% in the precision of contradiction detection.