[目的/意义] 通过对CORE论文关系发现过程及其服务的详细剖析，希望为我国开放获取知识库在论文内容的推荐和语义链接方面提供有力的参考和借鉴。[方法/过程] 从基于语义相似度的论文关联关系发现过程和基于论文关系的语义服务两方面进行分析。其中，基于语义相似度的论文关联关系发现过程包括元数据和全文内容收割、论文之间关系语义相似度计算两方面；基于发现的论文关联关系的语义服务包括论文推荐服务和关联开放数据服务。最后总结CORE对我国机构知识库的应用建议。[结果/结论] 研究发现，CORE系统通过现有OAI-PMH协议自动收割开放获取知识库中的元数据，并进一步提取元数据中URI字段，通过HTTP协议下载全文。基于发现的论文语义关系提供论文推荐服务和论文关联数据服务，使得第三方系统可以利用CORE数据集，这些都为我国开放获取知识库（如机构知识库、开放获取期刊）在论文关系的推荐和语义链接方面提供有力的参考。

[Purpose/significance] This paper dissects the process and services of article association discovery in Connecting Repositories, and hopes to provide powerful reference for the recommendation and semantic linking of the content of articles in Chinese open access repositories. [Method/process] This paper analyzed the discovery process of article association based on semantic similarity and the semantic services based on article association. The discovery process of article association based on semantic similarity included metadata and full-text content harvesting, and semantic similarity calculation of article association. The semantic service based on the discovery process of article association included the CORE recommendation service and the linked open data service. And this paper summarized the application suggestions of CORE to Chinese institutional repositories. [Result/conclusion] This paper finds CORE system automatically harvests the metadata of the open access repositories through the existing OAI-PMH protocol, and further extracts the URI fields from the metadata to download the full-text through the HTTP protocol. Furtherly, providing article recommendation services and services of data linked articles based on the discovery of article semantic association enables third-party systems to utilize CORE datasets, it provides a powerful reference in recommendation and semantic linking of article association for open access repositories (such as institutional repositories and open access journals) in China.