Host Organisation: Department of Informatics, Systems and Communication, University of Milano-Bicocca, Milan, Italy

The objective of contextual search is to leverage context information to improve the search outcome; moreover, the system should be able to detect the often complex and dynamic characteristics of the users' search tasks. Contextual search relies on the definition of user models, which may be formally represented in a variety of ways, such as bags of words or concepts, conceptual taxonomies or ontologies.

The project aims to analyze and evaluate the definition of neural network user models, in particular it has the following objectives: (1) to analyze the problem of representing the user's physical and cognitive context by means of neural network models; (2) to apply such model for contextual search and relevance assessment; (3) to instantiate the defined models on the considered applications and to define a suitable evaluation strategy.

The project encompasses issues related to both the definition of neural context models for IR, and the definition of suitable evaluation strategies for the various applications that will be considered.