论文部分内容阅读
提出一种合理运用大众标注数据自动构建资源导航体系的方法,首先使用聚类技术将标签分成主题相近的簇,然后在簇内按标签的泛指度和相似性构造标签树。将该方法用在公开的标签数据集SocialODP-2k9上进行实效评价,结果表明,本方法减少了以往标签树构造方法中“意义漂移”的缺陷,使组织结构有更好的语义一致性,且具有更好的导航效率,从而使本研究能适应信息组织与服务的实际需要。
This paper proposes a method to construct resource-based navigation system by using public annotation data reasonably. Firstly, the clustering technique is used to divide the tags into clusters with similar topics. Then the tag tree is constructed according to the degree and similarity of the tags in the cluster. The method was applied to the public tag data set SocialODP-2k9 for effectiveness evaluation. The results show that this method reduces the defects of “meaning shift ” in the past tag tree construction method and makes the organization structure have better semantic consistency , And has better navigation efficiency, so that this study can adapt to the actual needs of information organizations and services.