Sentiment Classification Of Commodity Evaluation Basrd On Parsing

With the development of e-commerce sites,more and more people are willing to make a survey on the internet before buying the product.but there is too many comments on the internet.It is impossible to look all of them.So we need the automatic identification of the comments.To know the emotional tendency of every comments,we take particularity of comments of products under the consideration and design one algorithm.The innovations of the algorithm is we used the parsing to get the objects and deal with the clause and the implied objects.Then We improve the algorithm according to short text on Micro blogging,as well as negative structure commodity evaluative statements,and we participate in the related evaluation,and achieved good results.This article using the syntactic analysis method to obtain the dependence structure of the statement,We design different algorithm to fit different structural of the sentence.At last we obtain the relationship between the evaluation object and the emotional tendency,the main steps are as follows: First,We extract the evaluation object of the evaluative statement,and design several filtering method to filter them and get the final collection of evaluation objects,and we allocate weights to them.Then,We chose the hotel,book and laptop to be the research objects. According to the particularity of products,We extract the object of the evaluation,make the allocation of weights,and then get the entire evaluation sentences emotional tendency and we came to the end of each evaluation for positive and negative results of the evaluation of the products.The result is good.Finally, for some of the special structure of the text,The previous algorithm was carefully the improved.We analysis the text from grammatical meaning and structure of short text,negative structure text.We proposed improvements of the method to each one and took part in related meetings and participated in the evaluation test,according to the release of the evaluation data, the result of the release,has a good performance.