Towards Improving Exploratory Faceted Search Results

Abstract

Finding information on the web, containing billions of documents, is becoming a serious issue. Search engines (SE) help users find handy information such as videos, text or pictures. For open-ended results, users need to perform queries in order to find enough information about their topic of interest. This can be difficult when huge numbers of relevant and irrelevant results are returned from the queries causing information overload. Certain SEs incorporating visualization allow easy understanding of the documents and often reduce the Information overload issues. Sorting of more relevant results from the less relevant ones is difficult since the former is insufficient. This is obvious when the Web grows and the information therein increases to become larger data repositories. The query and response (lookup) processes are often used by most current SEs. Exploratory Search (ES) can provide open ended search. Therefore, the overall search can be improved by applying the Faceted search (FS), which is one of the features of ES. Studies have mentioned that the overloaded data can be reduced when traditional SEs are replaced with the ES, through their features. The prototype design we are working on is based on the idea of ES engine that can be used to visualize the information when exploring the Web. This concept is aimed at supporting more active means of finding information through coordinated visualizations and reducing large amount of search efforts required per query. The prototype system was evaluated for performance and compared with CloudMining SE. The results show that our prototype gives average improvement of 20% relevant results when compared with CloudMining SE.