Understanding Sentiment Analysis

In the era of extracting insights from data, Sentiment Analysis is used to compute opinions, sentiments, views, etc. expressed in text format. Text polarity which recognizes sentiment inclination of text as positive or negative, Ranking (numbers in a range) the sentiment of the texts and Aspect based sentiment analysis which identify sentiments towards specific aspects in text are three broad divisions of problems in sentiment analysis. Supervised learning based sentiment analysis first trains and then tests the data. Unsupervised learning based sentiment analysis require sentiment dictionaries which can be created by lexicons or by a corpus based approach. Application of linguistic and statistical methods, training domain adapted models, aiming for aspect based sentiment analysis, identification of biometrics, images and sound as sources of sentiment data, etc. are some useful pieces of information shared in the blog. Contextual understanding, sentiment ambiguity and texts involving sarcasm and comparatives pose serious threats to performance of sentiment analysis system.