Applying Dependency Trees and Term Density for Answer Selection Reinforcement

Abstract

Abstract. This paper describes the experiments performed for the QA@CLEF-2006 within the joint participation of the eLing Division at VEng and the Language Technologies Laboratory at INAOE. The aim of these experiments was to observe and quantify the improvements in the final step of the Question Answering prototype when some syntactic features were included into the decision process. In order to reach this goal, a shallow approach to answer ranking based on the term density measure has been integrated into the weighting schema. This approach has shown an interesting improvement against the same prototype without this module. The paper discusses the results achieved, the conclusions and further directions within this research.

Information Retrieval with a Simplified Conceptual Graph-Like Representation

“…Rege et al [33] describe many forms of representing documents as graphs and Badia and Kantardzic [2] propose a methodology for construction of graphs via statistical learning. Graphs are widely used in natural language processing, for example, is question answering [26], text classification [4,10], named entity recognition [5], or information representation [3,31] (in combination with vector space techniques). In other cases the documents are represented by probabilistic functions [25] or composite function of probabilistic function on words [12,35].…”