Intelligent Question Answering System Of E-commerce Platform Based On Online Shopping Comments Key Technology Research

Under the background of the rapid development of artificial intelligence,intelligent natural language processing technology has made rapid development.The realization of efficient,comprehensive and more practical intelligent Q&A has become the key research direction of natural language intelligence.Among them,as one of the key application fields,with the help of the research results of natural language intelligent processing technology,the development and construction of more intelligent and practical intelligent question answering system for e-commerce field has attracted the attention of scholars and industry circles at home and abroad.In recent years,with the rapid development of e-commerce,the number of e-commerce platforms and users has increased rapidly,the variety of goods has become increasingly rich,and the explosive growth of commodity information.The problem of "information loss" in the purchase of commodities by the majority of online shopping consumers has become more and more serious,which leads to serious difficulties for consumers in the comparison and selection of commodities between different e-commerce platforms.On the one hand,from the perspective of market demand,due to the restriction of large user groups and limited customer service resources,the supply and demand of customer service is seriously unbalanced;On the other hand,from the perspective of technology research and development,different from other fields of Q&a system,the massive number of users of e-commerce platform and commodity content,as well as the different concerns and cognitive differences of different online shopping users for commodities,make the construction of intelligent Q&a system more challenging,many of which need to be further studied.Based on the above background,on the basis of summarizing and analyzing the main research results in related fields at home and abroad,this paper studies the key technologies of constructing the intelligent Q&a system in the field of e-commerce from the perspective of online shopping user comment analysis.The main research contents and achievements are as follows:(1)In the aspect of knowledge extraction of online shopping reviews,a named entity recognition method based on MA-BiLSTM-CRF is proposed and its effectiveness is verified by experiments.Firstly,the combination of word vector and part-of-speech vector is used as the input of BiLSTM to extract the global features of online shopping comment text.Then,the Multi Attention Mechanism is introduced to extract the final text features from the global features of the text.At last,we use CRF to identify the attribute and viewpoint knowledge in online shopping reviews.(2)In the aspect of named entity relation extraction,a CNN based entity relation extraction method is proposed and its effectiveness is verified by experiments.Firstly,the attribute entities are clustered by semi supervised clustering algorithm based on word vector.Then,the attribute entities are replaced according to the clustering results.Finally,CNN is used to extract the matching relationship between attribute class entity and viewpoint class entity.(3)In the aspect of knowledge base construction,a knowledge base construction method of Intelligent Question Answering System Based on knowledge map is proposed.In order to better represent and use named entities and entity relationships in online shopping reviews,this method firstly collates and extracts entities and entity relationships,then imports them into the graph data Neo4j,and finally constructs the knowledge base of Intelligent Question Answering System in the form of knowledge map.(4)In the aspect of Question answering system modeling,an intelligent Question answering processing flow based on knowledge map is designed,and a Question answering system modeling analysis and answer generation method that links the entities in the user's question text and the entities in the knowledge base accurately is proposed.This method is composed of question analysis module and answer generation module.First,the question analysis module uses entity recognition technology and entity link method to analyze the semantics of the questions and get the semantic information of the user's questions.Then,in the answer generation module,the semantic information obtained by the question analysis module is transformed into the knowledge map query language,and the answer of the question is obtained by querying in the knowledge map.Finally,the answers of the questions after word order processing are fed back to the users.