Algorithm Design And System Development Of E-commerce Q&A Based On Chinese Knowledge Graph

With the continuous generation of semantic data and large-scale accumulation of the Internet, the question and answer application because it can provide users with intelligent knowledge and services are favored. Typically, the question and answer application based on knowledge mapping is constantly produced. However, at present, most of the system and the research of these systems are in English, and the semantic representation of Chinese and English is very different, so there are still a lot of new challenges to the Chinese question answering system.In recent years, due to the rapid development of electronic commerce and the overall popular, a direct drive to the huge demand for customer service. In this paper, in view of the diversity and uncertainty of the user, to optimize the Chinese domain knowledge base is proposed an automatic question answering (ceqa) framework based on Chinese knowledge map, and try to improve the user experience and ease the pressure of artificial customer service from the point of view of the field of electronic commerce.The framework includes three parts:first, identify potential user entity in questions. For the special consideration of electronic commerce, this article uses the dictionary and the commodity characteristic, enhances to the brand, the model and the single product entity recognition performance. And integrated into CRFs to identify; secondly, semantic analysis on the user questions based on entity recognition. In view of the diversity and uncertainty of the user’s question, this paper introduces the dependent reduction algorithm, to reduce the semantic dependency analysis in the process of user questions noise, so as to improve the accuracy of semantic analysis. Finally, according to the question analysis results to the URL in the form of a link to the knowledge base. Diversity to overcome the user to express, namely multi word Italian phenomenon, the high similarity of word2vec link words, alleviate user representation of uncertainty.In order to verify the feasibility, efficiency and user friendliness of CEQA framework, this paper designs and implements a question answering system (CEQA-Live). The good operation of the real system provides a good environment for the performance of the framework, and it also proves the application potential of this study.