Research And Application Of Key Technologies Of Question Answering System Based On Knowledge Graph

Knowledge Graph can represent the relationship between entities and entities from a semantic perspective.It can formally describe the real world things and their relationship.With the development of automatic construction technology of knowledge Graph and the continuous progress of natural language processing technology,more and more intelligent question and answer systems based on knowledge Graph are applied in specific areas of automatic question and answer.In this paper,the key technologies in the construction of knowledge graph automation and the implementation of intelligent question and answer system are studied.The main work completed is as follows.(1)A data set denoising model based on Patterns and deep reinforcement learning is presented.Relationship extraction is one of the key tasks in the process of automatic construction of knowledge graph.Data can be automatically labeled by using remote monitoring methods,but there is a lot of noise data in the automatically labeled dataset.To solve this problem,this paper presents a data set denoising model based on Patterns and deep reinforcement learning.First,a pattern-based positive sample extraction algorithm is used.By labeling high frequency label data in the dataset,false positive examples in the positive sample dataset are eliminated.Secondly,Focal Loss is used to replace the traditional cross-entropy in the classification of convolution neural network,which effectively solves the imbalance between positive and negative sample data in the Relationship classification.Finally,the method of reinforcement learning is used to dynamically select whether to delete or retain the candidate instances of remote monitoring,and finally a high-quality remote monitoring dataset is obtained,which effectively reduces the noise data in the remote monitoring dataset.The noise removal performance of the model is verified by experimental comparison.(2)This paper designs and implements a Knowledge Q & a system of Luoyue Romance Based on knowledge map.In the field of war history and culture,this paper constructs a new entity relationship definition rule,which has a good reference for the future development of this field.In the system,the automatic construction of knowledge map in the field of war history and culture is completed through the dependency parsing model,and neo4 j is used as the database of the system;In the processing of users’ natural language questions,LTP model and similar dictionary are used to complete the semantic analysis of questions,which effectively improves the accuracy of answers.