Research on the construction of a knowledge graph for dual carbon policies based on natural language processing

LYU Tao; WANG Qingshan; ZHANG Ziyu; WU Yulei; ZHOU Zirou; WANG Luo

Abstract：The"dual carbon"policy is characterized by a high frequency of releases,broad scope,and complex,multifaceted content.The existing presentation methods are difficult to meet the needs of knowledge retrieval and internal analysis.This paper takes 2953 dual carbon policy texts as the data source and proposes a dual carbon policy knowledge graph construction method based on natural language processing.First,the knowledge graph model layer is constructed,and the dual carbon policy entities,attributes and relationships are defined.Then,the Text Rank keyword extraction,LDA topic modeling and other algorithms are used to extract policy entities,attributes and relationships to construct the knowledge graph data layer.Finally,the"entity,relationship,entity"triples are stored in the Neo4j graph database to form a dual carbon policy knowledge graph.The constructed knowledge graph contains 2048 entity nodes and 32336 relationships.The Cypher language enables the implementation of association queries and visualization of various fine-grained policy entities and their relationships.Additionally,it facilitates the extraction of key semantic information and the identification of policy hotspots within the dual carbon policy framework.It can also provide semantic enhancement functions for intelligent services,improve the efficiency of the dual carbon policy recommendation system and the accuracy of the policy question-answering system.