Research On Intelligent Question Answering Based On Knowledge Graph From The Perspective Of Knowledge Management

With the continuous promotion of China’s "education informatization" process,the integration of modern information technology and traditional teaching management methods has gradually become an important breakthrough point in the development of China’s education industry,creating rich teaching information resources and diverse knowledge service projects for the vast student population.However,the current massive teaching resources are characterized by decentralization and fragmentation,lack of integration and organization,making it difficult to achieve systematic knowledge management.At the same time,traditional knowledge service projects have problems such as low service efficiency and poor accuracy,and massive resources have not been properly and effectively managed and utilized,making it difficult to meet the needs of student users for high-quality information services.Therefore,how to provide a more rapid and high-quality course knowledge management and service content for the vast student population,and meet the learning needs of acquiring relevant course knowledge,has become an urgent problem to be solved.Knowledge Management,as a new management method that integrates modern information technology with classical management ideas,has emerged in the era of knowledge economy.It can achieve the construction of high-quality knowledge systems and the sharing of knowledge content.Applying the ideas and methods of knowledge management to course knowledge management tasks can effectively summarize and integrate scattered and redundant course resources and knowledge points in a certain course,improving the efficiency of knowledge acquisition.At the same time,various knowledge service contents derived from knowledge management ideas endow information resources with stronger innovation and sharing.Intelligent Q&A is a new and efficient knowledge service content that can automatically answer user questions;Knowledge graph is a mainstream knowledge management tool and method that can describe and integrate knowledge through a "graph network" structure.This article introduces knowledge graph technology into intelligent question answering tasks in the field of curriculum teaching.By studying the construction methods of course knowledge graph,a knowledge graph for the course "Python Language Programming" is constructed,and key models and algorithms in intelligent question answering tasks are studied,Finally,an intelligent Q&A mini program based on Python course knowledge graph was constructed,which achieved knowledge management tasks for a certain course content and provided high-quality knowledge service content for student users.The research work of this article mainly includes the following aspects:Firstly,the construction method of course knowledge graph.This article selects "Python Language Programming Course" and constructs a course knowledge graph using a top-down approach based on Python course materials,university MOOC website crawler data,and other data from different sources.Firstly,the ontology layer of the knowledge graph is modeled,and then the entity,relationship,and attribute triplet information required for constructing the knowledge graph is extracted from the raw data using a semi manual extraction method.Finally,the knowledge storage is implemented based on the Neo4 j graph database to construct the knowledge graph.Secondly,key models and algorithms for intelligent question answering.Decompose the intelligent question answering task in the field of Python courses into two subtasks: entity recognition and question classification.For entity recognition tasks,an entity recognition model based on Bi LSTM-ATT-CRF is constructed.By adding a reverse LSTM structure to the LSTM model,the model’s ability to extract context information of text sequences is improved.At the same time,ATT attention mechanism and CRF model are introduced to improve the model’s ability to extract key Semantic information,and at the same time,the final output tag values are constrained and standardized.For the question classification task,a question classification model based on BERT Text CNN is constructed.This model outputs a feature matrix containing more Semantic information through the BERT model,and uses the Text CNN model to extract the local features of short text questions,alleviating the problem of sparse short text vector features.After experimental testing,both models have achieved good performance in their respective tasks.Thirdly,intelligent question answering knowledge service content based on Python course knowledge graph.By designing and developing intelligent Q&A mini programs,we can automatically answer questions related to Python courses and provide efficient knowledge service content for student users.