Exploiting user feedback to improve quality of search results clustering
Inbeom Hwang, Minsuk Kahng, Sang‐goo Lee · 2011
Search result clustering provides an intuitive overview toward information contained in the search result. The goal of this research is to implement a clustering engine to provide search result clustering for various search tasks retrieving items, or objects whose contents do not contain descriptive text. Content-based similarity measures used for traditional clustering engines are not suitable for general measure, because of its domain-specific nature and lack of descriptiveness. To remedy the problems, we exploit user feedback information to measure similarity between items. As the first approach to use user feedback information to measure similarity between general items to cluster them, we explore similarity models and algorithms suitable for clustering.