This paper proposes a feature model different from bag-ofword models to analyze the sentiment of the text. The main idea of the method is improving the quality of prediction by combining a rulebased approach and the standard bag-of-words model. Results of the experiments with changing the subject, the size of reviews in data are shown. The hypothesis stating that it is better to use short message with the length of 1–2 sentences or tweets for calculation Delta TFIDF was tested.
CITATION STYLE
Samoylov, A. B. (2014). Evaluation of the Delta TF-IDF features for sentiment analysis. Communications in Computer and Information Science, 436, 207–212. https://doi.org/10.1007/978-3-319-12580-0_21
Mendeley helps you to discover research relevant for your work.