Research And Implementation Of Voluntary Service Recommendation System Based On Knowledge Graph

In the past ten years,the voluntary service has sprung up.With the promotion of voluntary service and the development of the Internet,a large number of voluntary service data has been generated in volunteer activities.How to deal with and utilize voluntary service data,guide the behavior of volunteer users,and provide valuable reference for participants in the volunteer cause,have become urgent problems to be solved.With the development of artificial intelligence and big data of Internet,the integration of "Internet+" voluntary service information has gradually developed.By analysis and research on massive voluntary service data,knowledge graph in the field of voluntary service is built.On the one hand,the complex structured voluntary service data is visually displayed in the form of a graph,and the Internet information is expressed in a form closer to human cognition,which facilitates better understanding and management of the data.On the other hand,realistic semantics and relationships are learned through computers,and intelligent services including personalized recommendation and intelligent retrieval are supplied to reference for the development of voluntary service.In this thesis,by mining voluntary service data and extracting entities and relationships,a knowledge graph is built in the voluntary service field for research on voluntary service recommendation.On the basis of representation learning model for knowledge graph,a recommendation algorithm is proposed,and voluntary service recommendation system is constructed based on the algorithm.The main work of the thesis includes:1.Based on big data of voluntary service,knowledge extraction is completed and the knowledge graph in the field of voluntary service is established.The analysis of the voluntary service data shows that there is an imbalance in the development of voluntary service.Establishing a knowledge graph in the field of voluntary service can help solve this problem.Firstly,manually define a hierarchical structure model based on the characteristics of voluntary service data,then extract entities,attributes,and relationships according to the model,and finally store the results of knowledge extraction in a combination of the MySQL database and the Neo4j database to complete the construction of the knowledge graph.2.Based on the knowledge graph in the field of voluntary service,representation learning algorithms for knowledge graph is studied,and the TransH-based recommendation method is proposed.Firstly,performances of the common knowledge graph representation learning models on the voluntary service data set are evaluated and the TransH algorithm performs best.Then,considering the problems faced by common recommendation algorithms,TransH-based recommendation algorithm is introduced,combining the representation learning model with classic collaborative filtering algorithm.By proposing concept of interest value,this algorithm is optimized to solve the cold start problem.Modeling the time decay function with time attribute of items helps the recommendation process to be closer to the real scene.Finally,the dense data set and the sparse data set are built to test the performance of the proposed algorithm.On the dense data set,the proposed algorithm improves the accuracy,recall,and coverage indicators by 29.91%,42.22%,and 23.81%respectively,compared to traditional item-based collaborative filtering algorithms.On sparse data sets the increment of three indicators are 38.78%,42.50%,and 24.55%.The result of experiment indicates that the proposed algorithm not only performs well in recommending accuracy and the ability to mine unpopular items,but also can alleviate the problem of data sparsity.3.Based on the knowledge graph and recommendation algorithm,a voluntary service recommendation system is implemented.This system provides data visualization service and project recommendation and retrieval service for volunteer users.Firstly,the three-layer architecture of the system is designed:data layer,business logic layer,and application service layer.The data layer uses MySQL and Neo4j databases.The business logic layer uses Spring framework and MyBatis.The application service layer is presented in the form of Web pages.Then the functional modules of the system are completed:volunteer registration and login module,volunteer graph information display module,voluntary service data display module,and voluntary project retrieval and recommendation module.The design scheme and process of each module are introduced,and the Web pages of each module is carried out.