Distributional Semantic Application to Information Retrieval from Large Textual Databases

Project Description

This research project concerns the development of semantic models for textual document retrieval systems. The models we are focusing on take place in framework based on a "distributional semantics" where semantic proximities are derived from co-frequency matrices computed on large textual corpora. The queries and documents are represented in an unified way as projections in a vector space of pertinent terms. Different similarity measures will be tested to characterize the proximity between queries and documents.

A software prototype, called D-SIR, has been implemented in order to validate the approach.