从社会性标签中进行语义关系抽

发布时间：2018-06-05 23:20

本文选题：关系抽取 + 标签　；参考：《现代图书情报技术》2009年03期

【摘要】：标签形式的社会性语义越来越占据主导地位,使元数据界在这种新形式的信息内容表达和检索方面面临机遇和挑战。其中,主要的挑战是与标签相关的语境信息的缺失。以Flickr标签为例,对如何利用社会性语义资源来丰富主题元数据进行了实验。实验过程包含4个步骤:收集Flickr标签样本;通过共有信息计算标签间的同现情况;通过Google检索结果来追踪标签对的语境信息;用自然语言处理和机器学习技术来抽取标签间的语义关系。本实验能够利用Google搜索结果构建语境库,并且以自然语言处理和机器学习算法对这些语句进行处理。这种新方法对于赋予标签对以一定语义关系有相当高的准确率。也探讨该方法在利用社会性语义丰富的主题元数据方面的意义。
[Abstract]:The social semantics of tag form are becoming more and more dominant, which makes the metadata community face opportunities and challenges in this new form of information content representation and retrieval. The main challenge is the absence of contextual information related to labels. Taking Flickr tags as an example, this paper makes an experiment on how to make use of social semantic resources to enrich topic metadata. The experiment consists of four steps: collecting samples of Flickr tags, calculating cooccurrence among tags through common information, tracing the contextual information of tag pairs through Google retrieval results, and analyzing the context of tag pairs. Natural language processing and machine learning techniques are used to extract semantic relationships between tags. This experiment can use Google search results to construct context library and process these sentences with natural language processing and machine learning algorithm. This new method has a high accuracy for giving tag pairs a semantic relationship. This paper also discusses the significance of this method in the use of social semantic rich topic metadata.
【作者单位】：美国雪城大学;华东师范大学信息学系;
【分类号】：G354

【相似文献】