SPARTE, a Test Suite for Recognising Textual Entailment in Spanish

摘要

The aim of Recognising Textual Entailment (RTE) is to determine whether the meaning of a text entails the meaning of another text named hypothesis. RTE systems can be applied to validate the answers of Question Answering (QA) systems. Once the answer to a question is given by the QA system, a hypothesis is built turning the question plus the answer into an affirmative form. If the text (a given document) entails this hypothesis, then the answer is expected to be correct. Thus, a RTE system becomes an Answer Validation system. Within this framework the first problem is to find collections for training and testing RTE systems. We present here the SPARTE corpus aimed at evaluating RTE systems in Spanish. The paper presents the methodology to build SPARTE from the Spanish QA assessments performed at the Cross-Language Evaluation Forum (CLEF) during the last three editions. The paper also describes the test suite and discusses the appropriate evaluation measures together with their baselines.