Improved Sentiment Classification From Meeting Transcripts
J.I. Sheeba, K. Vivekanandan · 2012
The web provides volumes of text-based data about customer preferences which are stored in online review websites, twitter, face book, blogs, etc. Sentiment classification has emerged as a method for mining opinions from such text archives and it uses machine learning methods combined with linguistic attributes or features in order to identify the sentiment polarity like positive, negative, and neutral for a particular document. Topic detection is also considered which helps in detecting the sentiment of each topic. In this proposed framework, it is able to identify sentiment and topic and also author classification from meeting transcripts. By improving its performance, it will incorporate the features of both JST and FRN method used here, it can identify sentiment and topic from the transcripts. In this Author classification, it is possible to identify both Author identification and Author characterization. Using SVM classifier, sentiments, topic and author classification are extracted .