Semantic inclination mining based on dependency grammar for Chinese BLOG

BLOG has made it possible to find out the opinions and experiences of those in the vast pool of people that are neither our personal acquaintances nor well-known. How to assess semantic inclination of BLOG contents and reviews automatically has attracted a lot of researchers concern. Due to Chinese expression is more complex than English, it is more difficult to assess semantic inclination of Chinese BLOG and accuracy is not high. This paper presents a continuous prior polarity algorithm. Our method reflects subtle change of sentiment in contrast to the previous studies which expressed sentiment polarity discretely. Furthermore, we propose a method to assess modified polarity based on Chinese dependency grammar. According to different dependency grammar relation, we can accurately identify subjective words and its modified polarity, then aggregate subjective words to predict sentence sentiment polarity. The experimental result illustrates that our method has higher accuracy compared with machine learning methods SVM and NB.