Research On Methods Of Short Text Classification For Customer Service Interaction Microblog

Text classification is an important research topic in the field of data mining. With twitter gradually occupied in foreign social field, more and more research began to focus on the microblog short text. Classification in microblog has an important significance in public opinion analysis, spam filtering, and other aspects of the microblog community. Contrasting to the domestic, Sina Weibo, as a representative, also began occupying people’s daily lives. However, due to the special nature of Chinese, classification for Chinese microblog short text also presented a greater challenge.Work of this paper is as follows:1. This paper investigates the relevant technologies of text classification, which include some data pre-process technologies,such as Chinese word segmentation and textual representation, feature selection techniques, as well as commonly used text classification algorithms and classification assessment methods;Putting forward some ideas to overcome the shortcomings of information gain feature selection method;2. Using LDA to put the microblog short text to the format of document-semantic distribution matrix.;3. Designing a classification method combined with the impoved information gain and LDA;Implementing a classification system based on customer service-oriented interactive microblog.In this paper, experiments are conducted on data of customer interaction microblog with the class label. Respectively, information gain and LDA as the comparisons, this design method on the classification accuracy obtained some improvement. The results show that the method in this paper applies to the classification for customer service-oriented interactive microblog.