Research And Implementation Of Water Affairs Knowledge Graph Construction And Question Answering Application System

With the development of the economy and society,the attention of the water industry continues to increase,and the technical demand of the industry also increases.At present,the internal system of Beijing’s water related departments includes a large amount of database data,and a lot of water related information is scattered on the network.As a result,water data cannot be well applied to water information management.Data and improve the efficiency of information services.In recent years,knowledge graphs have been favored by all walks of life,but there is relatively little research in the field of water affairs.Therefore,this paper builds a water affairs knowledge graph based on the current situation of water affairs data for knowledge representation and visualization.In addition,the question answering system is an effective way to simplify the acquisition of information.This article implements the knowledge question and answer application based on knowledge graphs to improve the efficiency of water affairs information retrieval.Based on these two analyses,the main research contents are as follows:First,through in-depth research on knowledge graphs,based on water data requirements and characteristics,this paper constructs water affairs knowledge graphs,including constructing top-level conceptual models of knowledge graphs,information extraction,and entity alignment,and visualizing knowledge graphs through Neo4 j platform.Among them,in the information extraction,in view of the fact that the entity information and semantic information in the water affairs text cannot be effectively obtained and used,a data extraction method based on the Pyltp named entity recognition model and Xpath statement is firstly proposed to achieve convenient and rapid extraction Entities Baidu Encyclopedia information;then combined with the pros and cons of the semantic role labeling model and dependent syntax analysis model for Chinese sentence analysis and the characteristics of Chinese sentences,a comprehensive grammatical analysis model was proposed to extract the semantic relationship between the subject-predicate-object triplet in the sentence,Contrast experiments and evaluations using water data have proved that the subject-predicate-object extraction method based on comprehensive grammatical analysis model proposed in this paper is more effective.Through the above two methods,the effective triplet information is extracted from the text.Entity alignment combines regular expression and edit distance algorithm to realize the mapping between entity and concept,and the comparison experiment with Jaccard algorithm proves the feasibility of this method.Secondly,by analyzing the characteristics of the question,this article designed the question analysis process.In order to improve the ability to analyze the question,according to the lack of existing water Chinese data sets and the lack of professional vocabulary in the word segmentation,a comparative analysis of various word segmentation Compared with the performance of the classifier,through analysis,the combination of Naive Bayes algorithm and Hanlp word classifier can improve the classification effect,and this method is verified by comparative experiments,which proves that the classifier constructed in this paper can adapt to the existing Water data.Finally,CQL query sentences and classification results are used to complete the search of answers from the knowledge base.Finally,in response to the needs of water users’ knowledge graph applications,we designed and developed a water affairs knowledge graph construction and question and answer application system,including overall architecture design,module function design and database design;using the Xitrum framework to complete the development of each system module and function,and Tests were conducted to verify the availability of the system.