Construction And Application Of Personnel Relationship Knowledge Graph For Social Governance

As China’s social governance enters a new stage,the government and public security departments need to better manage and utilize personnel relationship information in order to formulate policies,maintain social stability,and combat criminal activities.However,collaborative governance requires higher standards of data sharing and quality.Currently,there are many issues in data management and sharing,making it necessary to conduct correlation,authenticity,and completeness checks on massive amounts of personnel information data.To strengthen social governance and leverage intelligent support,this thesis introduces the knowledge graph method to improve data sharing and decision-making capabilities,and deeply mines the deep-level correlation information in social relationships for friendly and intuitive visualization.This study improves personnel decision-making capabilities by conducting in-depth mining analysis of social relationships to assist in pre-judgment and post-disposal of security incidents,and conducts empirical research targeting the Xinjiang Production and Construction Corps.The main work of this study includes:(1)Proposed Onto PI and OWL-based logical reasoning rules to create a structurally reasonable personnel organizational relationship knowledge graph,and conducted empirical research to improve data reusability,completeness,and credibility.(2)Designed the PIQA data evaluation method based on knowledge graph embedding model to provide early warning of anomalous information problems at the family structure level,reducing manual screening costs,improving personnel information data credibility and relevance,and ensuring high data quality.(3)Designed the end-to-end model GRPI under the Encoder-Decoder framework,combining overall graph structure information with local relationship information,to judge personnel identity.This method outperforms traditional machine learning methods and single graph mining methods in classification performance,achieving an accuracy of 83.6015 and an AUC score of 95.5519.The results show that the analytical methods combining graph data mining and knowledge representation learning can effectively analyze similar personnel and key personnel,which is important for downstream tasks.Finally,this thesis integrates the aforementioned methods into a personnel relationship analysis and judgment system,realizing four major functional modules: data integration,quality control,correlation analysis,and identity judgment.The research results will promote the application of knowledge graphs in personnel data governance and personnel relationship analysis,improve the ability of relevant departments to analyze and trace events,and achieve more effective management and decision-making.