Title :
A Random Walk Method for Sentiment Classification
Author :
Mingzhi, Cheng ; Yang, Xin ; Jingbing, Bao ; Cong, Wang ; Yixian, Yang
Author_Institution :
Inf. Security Center, Beijing Univ. of Posts & Telecommun., Beijing, China
Abstract :
Sentiment classification is very useful in many applications such as analysis of product reviews in e-business applications and media opinion mining. In this paper, a novel method to tag words sentiment automatically is proposed. In this method, a word association graph is firstly constructed from text corpus, i.e. product reviews, in which each node is a word and if there is an edge between two words, it means the two words co-occur in the same sentence. And then, with a random walk algorithm, the sentiment score is calculated for all the words in the graph at one time. To show the effectiveness of our method, the sentiment tagging results are then used for sentiment classification on real dataset. The experimental results show that the sentiment classification results with our method are better than the compared methods.
Keywords :
graph theory; natural language processing; pattern classification; text analysis; e-business applications; media opinion mining; product reviews; random walk method; real dataset classification; sentiment tagging; tag words sentiment; text corpus; text sentiment classification; word association graph; Conference management; Data engineering; Educational technology; Information management; Information technology; Laboratories; Learning systems; Machine learning; Support vector machine classification; Support vector machines; F-Measur; SO-PMI; SVM; random walk; sentiment classification;
Conference_Titel :
Future Information Technology and Management Engineering, 2009. FITME '09. Second International Conference on
Conference_Location :
Sanya
Print_ISBN :
978-1-4244-5339-9
DOI :
10.1109/FITME.2009.87
