This thesis is an attempt to contribute to the Information Retrieval field bydescribing methods to find, represent and use conceptual structures from natural language input. It is a pragmatic approach, not only since its main reasonis to practically improve retrieval systems, but also because it is inspired bypragmatism, the philosophical movement that claims that the meaning of con-ceptions is to be sought in their practical bearings [?]

This thesis answers the following research questions:

• Is it possible to use recent Natural Language Processing techniques toincrease the effectiveness of Information Retrieval?
• Is it possible to index a collection in such a way that it captures the mean-ing of its information items more accurate than word based models?