Comparing and evaluating the sentiment on newspaper articles: A preliminary experiment

Recent years have brought a symbolic growth in the volume of research in Sentiment Analysis, mostly on highly subjective text types like movie or product reviews. The main difference these texts have with news articles is that their target is apparently defined and unique across the text. Thence while dealing with news articles, we performed three subtasks namely identifying the target; separation of good and bad news content from the good and bad sentiment expressed on the target and analysis of clearly marked opinion that is expressed explicitly, not needing interpretation or the use of world knowledge. On concluding these tasks, we present our work on mining opinions about three different Indian political parties during elections in the year 2009. We built a Corpus of 689 opinion-rich instances from three different English dailies namely The Hindu, Times of India and Economic Times extracted from 02/ 01/ 2009 to 05/ 01/ 2009 (MM/ DD/ YY). In which (a) we tested the relative suitability of various sentiment analysis methods (both machine learning and lexical based) and (b) we attempted to separate positive or negative opinion from good or bad news. Evaluation includes comparison of three sentiment analysis methods (two machine learning based and one lexical based) and analyzing the choice of certain words used in political text which influence the Sentiments of public in polls. This preliminary experiment will benefit in predicting and forecasting the winning party in forthcoming Indian elections 2014.