基于语义表征的科技资源建模与发现方法

Semantic Representation-Based Modeling and Discovery of Scientific and Technological Resources

科技资源是科技创新的基础和保障,具有种类多、分布广、异构性强等特点。科技资源描述是对科技资源属性的人工标注和解释,已经成为科技资源元数据标准中的重要字段。标准不统一会造成描述歧义,给科技资源的信息集成、信息发现及资源利用造成了困难。本文提出了一种基于语义表征的科技资源建模与发现方法,旨在为有效解决科技资源描述歧义问题提供思路和方案。该方法通过选择合适的深度语义表征模型实现科技资源建模,并基于该模型生成资源向量实现深度语义匹配模型,最终实现具有良好消歧能力的资源发现功能。

Scientific and technological resources are the foundation of scientific and technological innovation,which have the characteristics of multiple types,wide distribution and strong heterogeneity. The description of science and technology resources is the manual labeling and interpretation of their attributes and it has become one of important fields in the metadata standard of science and technology resources. Specifically,the ambiguity caused by non-uniform standards has brought difficulties to information integration,information discovery and resource utilization. This paper proposes a method for the modeling and discovery of scientific and technological resources based on semantic representation,which aims to provide a model and method for effectively solving the ambiguity problem of scientific and technological resources. By selecting a suitable deep semantic representation model,this method realizes the modeling of scientific and technological resources based on which we proposed resource-vectors to realize a deep semantic matching model. Finally,this paper implements the resource discovery function with perfect disambiguation ability.