深度学习模型的矿业工程学科知识图谱构建

Construction of mining engineering disciplinary knowledge graph based on deep learning model

为解决矿业工程学科的知识组织中数据的异构性与稀疏性问题,构建一种可持续动态更新的矿业工程学科知识图谱。通过设计一种矿业工程学科的知识图谱本体模式,采用基于深度学习模型的知识抽取模型从多源异构数据源中获取领域知识,给出多特征的知识融合算法,实现了矿业工程学科知识图谱的知识扩展。结果表明,矿业工程学科知识图谱的构建方法的F_(1)值达到96.3%,融合方法的F_(1)值达到98.2%,可有效支持知识图谱动态持续构建以及学科热点等关联知识的可视化分析。

This paper aims to solve the problem of relevance and continuity in the knowledge organization of mining engineering disciplinary due to the heterogeneity and sparsity of data,and to construct a sustainable and dynamic updating knowledge graph of mining engineering disciplinary.The study involves designing an ontology mode of Knowledge graph of mining engineering disciplinary;combining the knowledge extraction model based on deep learning model to obtain domain knowledge from multi-source heterogeneous data sources;and proposing a multi feature knowledge fusion algorithm,as which realizes the knowledge expansion method of mining engineering discipline knowledge graph.The results show that the proposed methods with F_(1) 96.3% on knowledge graph construction,F_(1) 98.2% on knowledge fusion can dynamically update knowledge graph and visualize disciplinary hotspots analysis effectively.