在大数据和语义出版发展的背景下，针对科技文献知识组织与管理的实际需求，以语义出版为目标导向，探索科技文献的新资源描述框架，以期解决知识资源供给和科研用户需求之间的矛盾。本项目将语义出版与资源组织相结合，以知识供需两侧协同为切入点，重点研究科技文献的语义集成建模、深度语义标引、资源知识空间三方面内容。在语义集成建模研究中，对科技文献的多维知识特征进行统一描述和集成建模，构建基于知识元本体的多粒度知识本体集合，可弥补以术语或概念体系为基础进行语义建模的不足；在深度语义标引研究中，协同使用不同类型知识本体，对科技文献的不同层次知识特征进行深度语义标引，可改善以单一理论或视角进行浅层标引的现状；在资源知识空间研究中，构建由“资源-宏观-中观-亚微观-微观”五层次、多粒度知识单元组成的知识组织模型，并建立模型内外多元化的知识关联与链接，可突破图书情报领域中传统资源描述机制和知识组织方法的研究瓶颈。

In the development of big data and semantic publishing, with the goal of semantic publishing, we explore a new resource description framework for scientific literature in response to the actual needs of scientific literature knowledge organization and management to solve the contradiction between the supply of knowledge resources and the needs of scientific research users. This project combines semantic publishing with resource organization, and focus on the three aspects of semantic integration modeling, deep semantic indexing and resource knowledge space of scientific literature with the collaboration of knowledge supply and demand as the starting point. In the study of semantic integration modeling, we introduce a unified description and integrated modeling method for the multi-dimensional knowledge characteristics in scientific literature, and build a multi-granularity knowledge ontology collection based on knowledge elements ontology, which can cover the shortage of semantic modeling based on terminology or conceptual system. In the study of deep semantic indexing, we cooperate to use different types of ontology for deep semantic indexing of different levels of knowledge characteristics in scientific literature, which can improve the research status of shallow indexing with a single theory or perspective. In the study of resource knowledge space, we construct a knowledge organization model which consisting of five semantic levels and its multi-granularity knowledge units of resource, macro, meso, submicro, micro, and establish a diversified knowledge association and link inside and outside of the model, which can break through the research bottlenecks of traditional resource description mechanisms and knowledge organization methods in the field of library and information.