Sentiments analysis for textural resources is a challenging problem. This report focusses on helping students to build the ability of building sentiment identifying system using different approaches. This report consists of three parts. Part One is the carry-out of the assessment instructions, where a ready-to-execute Naïve Bayes model and a rule based model are illustrated. In Part Two, an improved rule-based system is implemented by the author. In Part Three, an innovative approach “Compositional Sentiment Analysis” is introduced, and is adopted to make comparison with the Naïve Bayes system in the assessment.
Please refer to Methodology.pdf for more information