RAPPORT — a Portuguese question-answering system

Abstract

We present a question answering system for Portuguese that depends on subject-predicate-object triples extracted from sentences in a corpus. It is supported by indices that store triples, related sentences and documents. The system processes the questions and retrieves answers based on the triples. For testing and evaluation purposes, we have used the CHAVE corpus, which has been used in multiple editions of CLEF. The questions from those editions were used to query and benchmark our system. In its current stage, the system has found the answer to 42% of the questions. This document describes the modules that compose the system and how they are combined, providing a brief analysis on them, and also some preliminary results, as well as some expectations regarding future work.