Design And Implementation Of Entity Extraction Module In A Public Security Inquiry Record Management System

As the most important part of the public security institutions in China,the public security organs are mainly responsible for two functions:one is maintaining social security,the other is investigating criminal cases.In the process of filing a case,a necessary link is to take notes of the suspects,reporters and witnesses.The record contains all kinds of information needed by the police,such as the identity information,family information,social information,crime tools,location,time and so on.These information play a great auxiliary role in the police’s handling of serial and parallel cases.To sum up the information,the traditional method is to rely on the manpower of the police,and the police manually extract the case record information.However,the efficiency of manual analysis is very low.If computer technology is used to extract the recorded information automatically,it will save a lot of time,so the public security needs a system that can help the police manage and analyze records,and use computers to help the police better solve cases.This paper investigates the products related to the public security inquiry record management system in the market,looks up the papers related to the design of the public security inquiry record management system,and investigates the named entity recognition method that can analyze the content of the record.According to the requirement of entity recognition and information extraction for the content of public security record,the outline design of entity recognition module in public security inquiry record management system and the detailed design based on Rule Template Method and neural network are completed.Among them,five template methods based on rules are place name entity recognition,time entity recognition and vehicle entity recognition,and extract theft case information and property loss case information.Neural network method is composed of bilstm and CRF,which can recognize place name entity,time entity and vehicle entity.Finally,the functional and non functional requirements of entity identification module are tested,and the accuracy,recall rate and F1 value of the two natural language processing methods are analyzed and summarized.This paper uses the above achievements,and implements the entity extraction module of the public security inquiry record management system by using SSM framework.This module is mainly used in online real-time calculus of input transcripts.The module has successfully fulfilled the established requirements and has a good performance in recognition accuracy.