Design And Implementation Of Knowledge Question Answering System Based On Event Graph Of Specific Scene

In recent years,the development of artificial intelligence has been very rapid,and it has become popular in people's production and life.In order to reduce labor costs,the customer service bar of many portals has launched an intelligent question and answer function,which provides users with answers and friendly conversations through the form of question and answer between users and intelligent machines.Among the many types of human knowledge,event logic is a very important and ubiquitous knowledge.Therefore,the event graph based on the logic of event relations was born from this,and the knowledge question and answer system based on the event graph can answer dynamic questions through the event logic.Event extraction and event relationship extraction are the cornerstones of the event graph,and its accuracy determines the quality of the event graph.This paper designs and implements a knowledge question answering system based on the event graph of a specific scenario.The main tasks include:(1)Starting from the task of Chinese event extraction,it solves some specific problems in Chinese event extraction,and proposes a new event extraction model ON-TLNN based on deep learning,and this model obtained the best result when compared with other event extraction models.(2)The entity relationship extraction method is introduced into the event relationship extraction task,and an ordered neuron model based on deep learning is used,combined with the attention mechanism,to complete the event relationship extraction task;and on this basis,ON-LSTM and the model combined with graph convolutional neural network finally achieved good results.(3)Completed the design and development of the knowledge question answering system based on the affair graph.First,the system is analyzed for overall requirements,system feasibility,functional requirements.;then the system is designed for architecture,functional module design,database design,and design and implementation of each module;finally,through functional testing and performance testing,it can be concluded that the overall performance of the system meets the expected performance standards.