VERB BASED MANIPURI SENTIMENT ANALYSIS

This paper deals about the sentiment analysis of the Manipuri article. The language is very highly agglutinative in Nature. The document files are the letters to the editor of few local daily newspapers. The text is processed for Part of Speech (POS) tagging using Conditional Random Field (CRF). The lexicon of verbs is modified with the sentiment polarity (Positive or Negative or Neutral) manually. With the POS tagger the verbs of each sentence are identified and the modified lexicon of verbs is used to notify the polarity of the sentiment in the sentence. The total number of polarity for each category that is positive, negative and neutral is counted separately. The highest total of the three is the deciding factor of the sentiment polarity of the document. The system shows a recall of 72.10%, a precision of 78.14% and a Fmeasure of 75.00%.