Session-based Suggestion of Topics for Exploratory Search

Exploratory information search can challenge users in the formulation of efficacious search queries to find the data they are interested in. Moreover, complex information spaces can disorient people, making it difficult to explore all the types of information relevant to their activities. In order address these issues, we propose a session-based concept suggestion model that, given the observed search queries, proposes context-dependent query expansions as a {\em ``you might also be interested in''} function. Our model can be applied to incrementally generate suggestions during the search sessions. This can be employed for query expansion, and in general to guide users in the exploration of the possibly complex space of information categories managed by an information system. Our model is based on the generation of a concept co-occurrence graph that describes how frequently concepts are searched together in sessions. Starting from an ontological domain representation, we generated the graph by analyzing the query log of a major search engine. Moreover, we identified clusters of ontology concepts which frequently co-occur in users' searches via community detection on the graph. An experiment carried out using the log provided satisfactory accuracy results.