Matrioshkas soft approaches to personalized web exploration

In this paper, we present the soft approaches and techniques developed within Matrioshka, a prototypal meta-search system aimed at providing personalized exploratory facilities, addressing the well known Ranked-List Problem. Matrioshka implements a novel interaction framework that provides tools for clustering documents retrieved by search engines, tools for exploring the content of clusters through the analysis of some cluster properties, tools for generating disambiguated queries from the clusters, and tools for combining the clusters to highlight their shared contents. All these tools are defined based on soft operations, in order to deal with intrinsic semantic ambiguity, imprecision and uncertainty of complex web searches. In this way, the user is supported in the deployment of complex web search exploratory activities.