WebRat 2002

WebRat- Interactive Visualisation and Analysis of Web Queries

Strategic Research Project

In the “WebRat” project we have combined the know-how and the exepriences acquired in other projects in the domain of Knowledge Retieval and Knowledge Visualisation to create a prototype tool for performing and refining Web search queries and and interactively visualising the result set. The retrieval and visualisation framework we have developed is capable of:

  • Querying different Web search engines and combining the search results.
  • Computing of similarities between Hits in a language independent manner.
  • Extracting thematic clusters from the textual information
  • Visualising of textual information based on thematic similarity
  • Metaphorical representation (landscape) of the combined search result sets
  • Providing of an interactive graphical user interface for exploration of the landscape Visualisation

After entering a search query the search results coming from different search engines are used to identify topical clusters and to produce a virtual thematic landscape. The generated landscape grows dynamically as more and more search results are gathered. The user can navigate the virtual map of serach results through panning and zooming. When athe user has identified a topic of interest the system can recommend additional search terms to refine and narrow down the original search query. With the help of WebRat the user can get an overview on the topical subjects covered by the search result set, learn about the vocabulary used in the topic of interest, and refine and narrow down the original serach query.

WebRat is conceived as an open framework, which can be extended to support different data sources (such as a database or a document management system). The front-end can also be easily re- configured to support different interactivity and visualisation styles.

Additionally a 1D visualisation component was developed with the aim of visualising hierarchically organised documents collections through a hierarchical bar diagram. The focus was on the separation of the presentation from the underlying data structure for the visualisation to be applicable in different contexts. The results will be employed in WebRat and IMB projects and, next year, will be integrated in the APA Intelligent Retrieval project.