Sentiment analysis based on clustering: a framework in improving accuracy and recognizing neutral opinions

摘要

Clustering-based sentiment analysis is a novel approach for analyzing opinions expressed in reviews, comments or blogs. In contrast to the two traditional mainstream approaches (supervised learning and symbolic techniques), the clustering-based approach is able to produce basically accurate analysis results without any human participation, linguist knowledge or training time.