Multidimensional Visualization and Navigation in Search Results

Abstract. Search engines traditionally index unstructured text and return ranked lists of documents that match a given query. As search engines functionally move in the direction of traditional databases and offer scalable and efficient handling also of structured textual and numerical data, this way of presenting results needs to be revisited and augmented with alternative presentation modes. This paper presents a multidimensional visualization scheme with navigational capabilities, built on top of a data aggregation framework implemented in a state-of-the-art search engine. The proposed visualization scheme is designed to provide analytic understanding of the result set also for ad hoc queries, and correlates user-specified dimensions or attributes of the documents. By employing advanced linguistic and entity extraction techniques, the scheme can also be applied in situations where the original documents consist of unstructured text.

• Type: Conference
• Year: 2004
• Where: KES
• Authors: Will Archer Arentz, Aleksander Øhrn
• Keywords: KES 2004, Search Engine, Unstructured Text, Visualization Scheme