FEATURE BASED ANALYSIS IN OPINION MINING VIA DOMAIN RELEVANCE

Abstract

This paper helps in identifying the features from online reviews by applying feature filtering criterion. Existing opinion feature extraction techniques are mainly based on mining patterns from a single review corpus which is most of the times dependent review corpus. Identifying candidate features which are from both corpora i.e. domain dependent and domain independent, this is captured by a measure called Domain relevance. Features extracted from this are relevant to a domain. For each extracted candidate feature its respective Intrinsic Domain Relevance and Extrinsic Domain Relevance values are estimated. These values are compared with threshold and are identified as best candidate features.