Study On Multivariate Quantification For Consumption Sentiment Classification Based On Three-dimensional Coordinate

Consumption sentiment of comments has an important reference value in buyers" purchase decision and sellers improving the way of business. Aiming at the problem that existing methods of consumption sentiment classification is two classes and the comment rating are labeled falsely, a multivariate quantification consumption sentiment classification method was produced. Firstly, construct the method of choosing Chinese sentiment phrases according to Chinese grammatical structure, and extract sentiment phrases of comment sets; Secondly, calculate the seed value of phrases based on the seed phrase set of consumption sentiment selection algorithm, and choose the seed phrase sets of different consumption sentiment classes; Thirdly, use comments set in place of search engine to search and calculate the co-occurrence probability between sentiment phrase and seed phrase, as well as get sentiment values of different classes; At last, three-dimensional coordinate is constructed, according to multivariate quantification consumption sentiment classification algorithm, locate the comment to the coordinate, realize the quantification of comment and classify the comment. Carried on the experiment of three comment sets, extracted the sentiment phrases according to the rule, and chose the seed phrase sets based on the seed phrase set of consumption sentiment selection algorithm, then mapped comment into the three-dimensional coordinate based on multivariate quantification consumption sentiment classification algorithm, which realized consumption sentiment classification of comments, meanwhile compared with the original PMI algorithm, which validated the effectiveness of this method in the Chinese consumption sentiment classification.