tailieunhanh - Báo cáo khoa học: "Query-based sentence fusion is better defined and leads to more preferred results than generic sentence fusion∗"

We show that question-based sentence fusion is a better defined task than generic sentence fusion (Q-based fusions are shorter, display less variety in length, yield more identical results and have higher normalized Rouge scores). Moreover, we show that in a QA setting, participants strongly prefer Q-based fusions over generic ones, and have a preference for union over intersection fusions. | Query-based sentence fusion is better defined and leads to more preferred results than generic sentence fusion Emiel Krahmer Tilburg University Tilburg The Netherlands Erwin Marsi Tilburg University Tilburg The Netherlands Paul van Pelt Tilburg University Tilburg The Netherlands Abstract We show that question-based sentence fusion is a better defined task than generic sentence fusion Q-based fusions are shorter display less variety in length yield more identical results and have higher normalized Rouge scores . Moreover we show that in a QA setting participants strongly prefer Q-based fusions over generic ones and have a preference for union over intersection fusions. 1 Introduction Sentence fusion is a text-to-text generation application which given two related sentences outputs a single sentence expressing the information shared by the two input sentences Barzilay and McKeown 2005 . Consider for example the following pair of sentences 1 1 Posttraumatic stress disorder PTSD is a psychological disorder which is classified as an anxiety disorder in the DSM-IV. 2 Posttraumatic stress disorder abbrev. PTSD is a psychological disorder caused by a mental trauma also called psychotrauma that can develop after exposure to a terrifying event. Thanks are due to Ed Hovy for discussions on the Rouge metrics and to Carel van Wijk for statistical advice. The dataset described in this paper 2200 fusions of pairs of sentences is available upon request. This research was carried out within the Deaso project http . 1All examples are English translations of Dutch originals. Fusing these two sentences with the strategy described by Barzilay and McKeown based on aligning and fusing the respective dependency trees would result in a sentence like 3 . 3 Posttraumatic stress disorder PTSD is a psychological disorder. Barzilay and McKeown 2005 argue convincingly that employing such a fusion strategy in a multidocument