基于全文知识网络的学术资源关联发现实践

Discover Linked Relations among Academic Resources Based on Full-Text Knowledge Network

【目的/意义】通过对学术资源进行深度挖掘与语义化组织,实现学术资源及其内部知识之间的关联发现。【方法/过程】本文提出基于全文知识网络的学术资源关联发现方法,设计了全文知识网络的模型和构建流程,以Pubmed Central数据库中拟南芥(Arabidopsis)相关的520篇期刊论文全文数据为实验对象,通过全文解析和挖掘将其分解为细粒度的知识,形成全文知识网络。然后利用SPARQL查询和RelFinder可视化工具从数字资源层、知识单元层和知识对象层三个层次开展关联发现实验。【结果/结论】本文构建全文知识网络对学术资源进行细粒度组织和挖掘,有助于发现不同学术资源及其内部知识之间的潜在关联,对学术资源的深度利用具有重要的意义。【创新/局限】本文创新之处在于通过构建全文知识网络对学术资源进行细粒度揭示和组织并进一步发现潜在关联,局限在于尚未开展大规模应用实践。

【Purpose/significance】Discover the linked relations among academic resources and their internal knowledge by structuring and organizing academic resources semantically.【Method/process】This paper proposes a method to discover the linked relations of academic resources based on full-text knowledge network, and designs the model and construction process of full-text knowledge network. 520 full-text articles about Arabidopsis are downloaded from the Pubmed Central database. After parsing, knowledge extraction,these articles are described as fine-grained knowledge and constructed into full-text knowledge network. Then SPARQL query and RelFinder visualization tool are used to discover the linked relations between entities in three layers(academic resource layer, knowledge unit and knowledge object layer) of the knowledge network.【Result/conclusion】 This paper constructs a knowledge network to fine-grained organize and reveal academic resources, which is helpful to discover the potential relations between academic resources.It is of important significance for the deep utilization of academic resources.【Innovation/limitation】The innovation of this paper lies in the fine-grained representation, organization and further relation discovery of academic resources through the construction of full-text knowledge network, and the limitation is that large-scale application practice has not been carried out.