基于概念空间的文本语义索引

发布时间：2018-11-15 09:13

【摘要】：正 1 引言据统计,在现今的联机存储信息中,80％以上的信息以文本的形式存在。信息的多元化、复杂化,致使信息的自动索引成为急需解决的问题。本文研究的内容是建立一个基于概念空间的文本语义索引。目前的文本索引都是建立在文本空间,或关键词空间上的,而建立在概念空间上的索引具有条理清晰、人机界面友好、符合通常检索习惯等许多优势,这也是文本语义索引发展的方向。另外.在建立文本索引的过程中,国内外大多使用Hopfield神经网络联想的方法,本文首次使用直接聚类法代替了Hopfield神经网络联想功能,这样使得索引具有很好的可扩展性。基于语义关联度的文本索引可以广泛应用于Internet搜索引擎、数字图书馆、电子商务等众多领域中。建立文本索引的过程主要有以下几部分: 1)对文档分类,建立文档的概念空间,在概念空间的层次上组织文档并确定文档中出现的关键词。
[Abstract]:According to statistics, more than 80% of the online storage information exists in the form of text. The diversification and complexity of information makes automatic indexing of information an urgent problem. The content of this paper is to build a semantic index of text based on concept space. At present, all text indexes are built on the text space or keyword space, but the index based on the concept space has many advantages, such as clear structure, friendly man-machine interface, consistent with the common retrieval habits and so on. This is also the direction of text semantic index. In addition. In the process of establishing text index, most of the methods of Hopfield neural network association are used at home and abroad. In this paper, the direct clustering method is used to replace the Hopfield neural network association function for the first time, which makes the index have good expansibility. Text indexing based on semantic correlation can be widely used in many fields, such as Internet search engine, digital library, electronic commerce and so on. The process of establishing text index includes the following parts: 1) classifying documents, establishing concept space of documents, organizing documents at the level of concept space and determining the keywords appearing in documents.
【作者单位】：中国科技大学研究生院计算机学部中科院计算技术研究所智能信息处理开放实验室中科院计算技术研究所智能信息处理开放实验室中科院计算技术研究所智能信息处理开放实验室
【基金】：国家自然科学基金(课题号60073019、69803010)
【分类号】：TP391.1

【相似文献】