In this talk, we are going to introduce the main ideas behind our extraction information system. The proposed model has two relevant and novel characteristics:

a) Take into account uncertain, modal, temporal and ontological information already present in the source of extraction. So, it looks to keep more rich information in the outcome of the extraction process
b) Incorporate inference engines into IES. These inference engines should be able to deal with uncertain, modal temporal and ontological information, in order to obtain explicit information that might be deducted from extracted information.
 
In order to achieve these goals we have taken an interdisciplinary approach using Natural Language Processing and Computational Logic techniques.
 
Using NLP techniques, we have developed algorithms to extract not only participants and eventualities but also their temporal and modal properties. We follow an already established methodology, using corpus analysis, cascaded finite state automata and deterministic grammars, machine learning approaches, in a cycle that involves training evaluation and refinement.
 
From a Computational Logic point of view, we have used multi-valued propositional logics and/or modal-temporal logics. We have developed specific applications (mostly automatic provers and model checkers) in order to be able to reason using these logics. Our goal is to obtain more expressive power than the classic propositional logic without reaching the computational complexity of first order logic.