| 任超,杨孟辉,许嘉元.基于引用关系的政策与论文关联图谱构建与应用研究[J].数字图书馆论坛,2025,21(8):76~84 |
| 基于引用关系的政策与论文关联图谱构建与应用研究 |
| Construction and Application of Policy-Paper Association Graph Based on Citation Relationships |
| 投稿时间：2025-06-17 |
| DOI：10.3772/j.issn.1673-2286.2025.08.008 |
| 中文关键词: 引文分析；知识图谱；政策引用论文；论文推荐；循证决策 |
| 英文关键词: Citation Analysis; Knowledge Graph; Policy-Citing Paper; Paper Recommendation; Evidence-Based Policy Making |
| 基金项目:本研究得到国家社会科学基金项目“基于图书全内容的知识发现与智能服务研究”（编号：22BTQ068）、宁波工程学院科研启动基金项目“多源数据驱动的科学传播生态系统研究”（编号：24KQ050）资助。 |
| 作者 | 单位 | | 任超 | 宁波工程学院人文与艺术学院 | | 杨孟辉 | 中国人民大学信息资源管理学院 | | 许嘉元 | 中国科学院文献情报中心 |
|
| 摘要点击次数: 30 |
| 全文下载次数: 40 |
| 中文摘要: |
| 当前，科学研究成果在政策制定中的作用日益凸显，但二者之间的隐性关联尚不明晰且挖掘难度较大，亟须构建系统性工具以帮助政策制定者准确有效地选择科学证据。从政策引用的视角出发，旨在基于大规模数据构建政策与论文关联图谱，在此基础上为政策制定提供科学论文推荐。在图谱构建阶段，以Overton和OpenAlex两大数据库为基础数据源，设计并构建融合多类型、多层级实体的知识图谱模式层（包括9种实体与5种关系），并采用自顶向下的知识图谱构建策略，抽取并生成47 327 880个语义三元组，存储于Neo4j图数据库中，实现高效查询与可视化支持。在图谱应用阶段，使用6种知识图谱推理技术，结合5种评价指标对结果进行评价，结果表明所提方法能够更为准确和高效地为政策制定推荐科学论文。研究不仅为实现基于知识图谱技术的政府决策提供了可行框架和具体建议，也为政策智能化支持工具的探索提供了有价值的理论与实践基础。 |
| 英文摘要: |
| While the role of scientific research in policymaking is increasingly significant, the latent associations between them remain obscure and are difficult to mine. This creates an urgent need for systematic tools that can aid policymakers in selecting scientific evidence accurately and effectively. Adopting the perspective of policy citations, this paper aims to construct a knowledge graph of policy-paper linkages from large-scale data to provide scientific paper recommendations for policy formulation. In the construction phase, we utilize the Overton and OpenAlex databases to design a schema layer for the knowledge graph, integrating multi-type and multi-level entities (nine entity types and five relation types). A top-down construction strategy is then employed to extract and generate 47?327?880 semantic triples, which are subsequently stored in Neo4j graph database to enable efficient querying and visualization. In the application phase, we apply six distinct knowledge graph reasoning techniques and evaluate their performance using five evaluation metrics. The results demonstrate that the proposed method can recommend scientific papers for policymaking with high accuracy and efficiency. This study not only provides a feasible framework and concrete suggestions for knowledge graph-driven government decision-making, but also establishes a valuable theoretical and practical foundation for the development of intelligent policy support tools. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
