Design And Implementation Of Knowledge Map System For The Urban Administrative And Law Enforcement

With the acceleration of China’s urbanization process and the continuous improvement of the law enforcement system,the pressure of urban management and law enforcement is increasing year by year,which puts forward higher requirements for the efficiency and quality of law enforcement personnel in handling cases.With the continuous progress of artificial intelligence,urban administrative and law enforcement agencies seek to build an intelligent and guided urban management pattern.As a structured semantic network,knowledge atlas has the ability of systematic semantic processing and open interconnection,which can well organize the knowledge information in different structure data and provide systematic knowledge services for urban governance.Establish a knowledge map in the field of urban management and law enforcement,guide urban managers to make a more fair and reasonable punishment system through intelligent event analysis and handling suggestions,improve the efficiency of law enforcement,and accelerate the pace of intelligent urban construction.Information extraction is the core of knowledge map construction,and the result of knowledge information extraction directly affects the quality of domain knowledge base.In order to ensure the quality of information extraction,it is necessary to study the existing entity relationship extraction methods.At present,the entity relationship extraction task mainly uses the method based on deep learning.On the premise of in-depth study of different model structures,this paper proposes a feature enhancement attention model based on segmented convolution neural network.Based on the segmented convolution neural network,this paper makes three improvements in semantic information representation and key feature extraction,1)introduces the pre-training model of Chinese word granularity to generate dynamic word vectors based on context semantics,2)word information is integrated with entity relative distance and part of speech information to enhance sentence feature information,3)proposes a sentence-level attention mechanism based on relational category keywords,Data noise reduction processing and strengthening the learning of relationship key information.Comparative experiments were carried out on the baseline model and the improved points to verify the effectiveness of the improved method.Compared with the PCNN model,the overall accuracy of the new model in relation extraction performance of multiclass data sets was improved by about 7.1%.Based on the proposed relationship extraction model and the implementation method of entity recognition task,this paper proposes a complete set of knowledge extraction task solutions for the field of urban management and law enforcement.Based on the knowledge extraction method,the knowledge atlas system of B/S structure is designed to build the domain knowledge base by completing the knowledge extraction of unstructured text,and provide the knowledge query function to support the construction of intelligent urban management pattern.