Adaptive Tag Extraction For The Custermer Service Micro-Blog Automatic Question And Answering System

With the development of Internet, online search has already become a new way to get information. Users hope to get their interested information quickly and easily. While search engine returns result in the form of web page list, which requires users to find the answer by themselves, so the search engine reduces efficiency and also affects users’experience. Automatic question and answering system which returns the accurate and brief answers to users and it allows users to ask question with the natural language. Automatic question and answering system covers the shortage of search engine and improves the convenience of users. The scope of application of automatic answering system is widespread, it is very important in the field of natural language processing research direction.In this paper, the key technologies about the automatic answering system have been researched and improved, and have achieved some achievements, the main work includes the following aspects:(1) Proposed an improved clustering algorithm which combines Canopy clustering algorithm and LDA topic model algorithm.(2) Proposed an improved tag extraction algorithm which is based on Text Rank tag extraction algorithm, which considered the figure structural information and semantic information between words which is computed by Word2vec.(3) Combined text clustering algorithm and tag extraction algorithm to achieve the function of adaptive tag extraction.(4) Designed the adaptive tag extraction for the customer service micro-blog automatic question and answering system, which combines the characteristics of user’s questions and customer service’s answers on "weibo.com".Experiments and comparative analyses prove that the improved clustering algorithm reduces the training time and improves the clustering effect, and the improved tag extraction algorithm get better effect than traditional tag extraction algorithms.Through the test, the system, that adaptive tag extraction for the customer service micro-blog automatic question and answering system, enhances the usability of knowledge base and the accuracy is up to70%. The system’s mean response time is less than one second. So the system meets the basic needs of users, and can help enterprises to complete customer service work efficiently.