从计算角度看大规模数据中的知识组织

Knowledge Organization in Large-scale Datasets from Computational Perspective

大数据时代的到来促使各种大规模数据集不断涌现,如何组织其中的知识信息以提供内容更加丰富、功能更加强大的知识服务成为当前的研究热点。文章根据大规模数据中知识内容多元化、结构网络化、源数据非结构化以及状态频繁演化等特性,从计算角度对其知识组织的重点问题进行了探讨。文章认为,知识组织需要适应当前以知识复用、发现和增值为核心的知识计算服务的需求,其重点在于底层知识信息的组织管理并受到知识持续演化的重要影响。文章提出了以语义数据管理为基础进行知识组织的观点,并据此对大规模数据中知识组织的核心问题(包括语义数据模型、知识表示、知识计算等方面)进行了剖析,论述了各相关方向亟待解决的理论问题和未来可能出现的发展趋势。

Various kinds of large-scale datasets spring up in the big data era, which makes it a hot topic for organizing the knowledge embedded in datasets to provide more abundant and powerful knowledge services. In this paper, we explore the features of the knowledge in large-scale datasets, such as content diversity, network-alike structure, unstructured data sources and ongoing evolution, and investigate the important issues of knowledge organization from a computational perspective. We assume that knowledge organization methodology should meet the requirements of computational knowledge services which focus on the utilization, discovery and increment of knowledge. The keypoint here is that the organization and management of knowledge information is substantially affected by the ongoing knowledge evolution. Consequently, we point out that the knowledge organization should be built upon proper semantic data management. Furthermore, the key issues of knowledge organization for large-scale datasets have been discussed. The relevant theoretical topics, such as semantic data modeling, knowledge representation and knowledge computation are explored to address the urgent problems and the possible emerging trends of related fields are also mentioned.