Research And Application Of Intelligent Holographic Archive System Based On Knowledge Graph

"Technological Empowerment of Police" is one of the important themes for China’s public security work to be promoted and transformed.In the current new era of social development and informatization,the public security department needs to continuously introduce emerging technologies in order to take the initiative in new environments,solve new contradictions and problems.Currently,the police informationization archive system mainly stays in the stage of using existing structured data,and there are great difficulties in data governance and processing for other types of data.Current data governance and archiving still rely mainly on manual labor,which makes it urgent to automate and intelligentize the processing of police data.In the process of intelligent archive generation,there are many difficulties,including identifying entities and extracting relationships between entities from unstructured text data,and how to govern and process disorganized information to obtain structured data that can be used for intelligent archives.This paper is based on knowledge graph technology and has researched and designed an intelligent holographic archive system based on knowledge graph.The system not only covers the basic functions of traditional police archive systems such as management and queries,but also can construct a holographic visualization archive based on multiple information dimensions such as personnel,vehicles,and cases.The system incorporates knowledge graph technology and uses artificial intelligence technology for knowledge generation,fusion,and processing,successfully achieving the automation and intelligentization of police data processing,and solving the problem of manual data processing on personnel quality and knowledge.This research mainly focuses on the following aspects:(1)Using artificial intelligence models such as bidirectional recurrent neural network and conditional random field algorithm,combined with rule-based and dictionary-based methods,to train an entity recognition model that is more suitable for Chinese police record data with a small amount of labeled data.The model’s overall recognition rate exceeds 95%,which is higher and more stable than traditional named entity recognition algorithms,providing accurate and efficient entity recognition results for knowledge graph construction and laying good data foundation for step(2).(2)Using bidirectional GRU(RNN)algorithm and rule-based and dictionary-based methods to recognize and classify entity relationships and train a relationship classification model that is more suitable for Chinese police record data,obtaining relationships between persons and persons,and persons and entities.The model uses specific entity results provided by step(1)named entity recognition algorithm for relationship extraction.It has a recognition rate of over 90%for relationship types unique to record data,efficiently identifying the relationships between individuals and accurately identifying suspects and related individuals in cases.(3)Designing and implementing a knowledge graph construction algorithm for Chinese police record data,which integrates and processes the entity information identified in step(1)and the relationship generated in step(2)to construct a knowledge graph.The knowledge graph is stored in multiple databases,serving as an important data source for police situation analysis and presented intuitively through the holographic archive system.