Summary: TALP­QA System for Spanish at CLEF­2004:
Structural and Hierarchical Relaxation over
Semantic Constraints
Alicia Ageno 1 , Daniel Ferr'es 1 , Edgar Gonz'alez 1 , Samir Kanaan 1 ,
Horacio Rodr'iguez 1 , Mihai Surdeanu 1 , and Jordi Turmo 1
TALP Research Center, Universitat Polit`ecnica de Catalunya,
Jordi Girona 1­3, 08043 Barcelona, Spain
fageno,dferres,egonzalez,skanaan,horacio,surdeanu,turmog@lsi.upc.es
http://www.lsi.upc.es/~nlp
Abstract. This paper describes TALP­QA, a multilingual open­domain
Question Answering (QA) system that treats both factoid and defini­
tional questions. The system is described and evaluated in the context
of our participation in the CLEF­2004 Spanish Monolingual QA task.
Our approach to solve factoid questions is to build a semantic repre­
sentation of the questions and the sentences in the passages retrieved
for each question. A set of Semantic Constraints (SC) are extracted for
each question. An answer extraction algorithm extracts and ranks sen­
tences that satisfy the SCs of the question. If matches are not possible
the algorithm relaxes the SCs structurally (removing constraints) and/or
hierarchically (abstracting the constraints using a taxonomy).

Source: Ageno, Alicia - Departament of Llenguatges i Sistemes Informátics, Universitat Politècnica de Catalunya

Collections: Computer Technologies and Information Sciences