知识产权法律法规知识图谱

知识产权法律法规知识图谱数据集包括615个高度结构化、知识化的知识图谱文件，包括全部中国现行的知识产权法律法规、司法解释、裁判规则、部门规章的知识三元组，可以高效用于知识检索、知识推送、知识支撑，命名的逻辑为语义逻辑聚类和专家标注后的法律规范类别，数据总量为411MB，可以高效用于知识检索、知识推送、知识支撑。技术上，以前述分片数据为输入，本研究采用基于大语言模型（LLM）的生成式方法，经由实体-实体、实体-事件、事件-事件三阶段递进式抽取，构建了实例级三元组池；进而，通过一种上下文增强的模式归纳算法，自动化地实现了节点与关系的概念化抽象，形成了蕴含丰富语义模式的知识图谱。The Intellectual Property Laws and Regulations Knowledge Graph Dataset consists of 615 highly structured and knowledge-encoded knowledge graph files, which cover all currently valid Chinese intellectual property laws and regulations, judicial interpretations, adjudicative rules, and departmental regulations in the form of knowledge triples. It can be efficiently used for knowledge retrieval, knowledge push, and knowledge support. The naming logic follows the legal norm categories derived from semantic logic clustering and expert annotation. With a total data volume of 411 MB, the dataset can also be efficiently utilized for knowledge retrieval, knowledge push, and knowledge support. Technically, taking the aforementioned segmented data as input, this study adopts a generative method based on Large Language Model (LLM), and constructs an instance-level triple pool via a three-stage progressive extraction process covering entity-entity, entity-event, and event-event relations. Furthermore, through a context-enhanced pattern induction algorithm, it automatically realizes the conceptual abstraction of nodes and relationships, forming a knowledge graph rich in semantic patterns.

搜集汇总

背景与挑战

背景概述

该数据集是一个包含615个知识图谱文件的知识图谱资源，涵盖中国现行知识产权法律法规、司法解释等高度结构化的知识三元组，数据总量约411MB。它采用基于大语言模型的生成式方法构建，适用于知识检索、推送与支撑等场景。