An optimized rule based approach to extract relevant features for sentiment mining

摘要

The popularity of internet has led to the generation of huge volume of unstructured data in the form of reviews, blogs etc. If processed efficiently, this data can act as potential goldmine for discovering knowledge which would prove very useful for both customers and market researchers. So the task of feature extraction in Opinion Summarization plays a very crucial role. In this research direction of extracting features, the paper proposes a framework model and an automated rule generation algorithm for extracting features of relevance. The technique proposed is an unsupervised and domain independent approach. The performance of the proposed technique was examined on 5 different data sets that were publically available and on two others that were manually crawled and tagged. The result demonstrated the percentage of features that were matching with the ones in the golden data set was more than 70% and reduction in the feature space was also considerably high.