Exploiting Discourse Analysis for Article-Wide Temporal Classification

Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing, pages 12–23, Seattle, Washington, USA, 18-21 October 2013. c⃝2013 Association for Computational Linguistics Exploiting Discourse Analysis for Article-Wide Temporal Classification Jun-Ping Ng1, Min-Yen Kan1,2, Ziheng Lin3, Wei Feng4, Bin Chen5, Jian Su5, Chew-Lim Tan1 1School of Computing, National University of Singapore, Singapore 2Interactive and Digital Media Institute, National University of Singapore, Singapore 3Research & Innovation, SAP Asia Pte Ltd, Singapore 4Department of Computer Science, University of Toronto, Canada 5Institute for Infocomm Research, Singapore junping@comp.nus.edu.sg Abstract In this paper we classify the temporal relations between pairs of events on an article-wide ba- sis. This is in contrast to much of the exist- ing literature which focuses on just event pairs which are found within the same or adjacent sentences. To achieve this, we leverage on dis- course analysis as we believe that it provides more useful semantic information than typical lexico-syntactic features. We propose the use of several discourse analysis frameworks, in- cluding 1) Rhetorical Structure Theory (RST), 2) PDTB-styled discourse relations, and 3) topical text segmentation. We explain how features derived from these frameworks can be effectively used with support vector machines (SVM) paired with convolution kernels. Ex- periments show that our proposal is effective in improving on the state-of-the-art signifi- cantly by as much as 16% in terms of F1, even if we only adopt less-than-perfect automatic discourse analyzers and parsers. Making use of more accurate discourse analysis can fur- ther boost gains to 35%. 1 Introduction A good amount of research had been invested in un- derstanding temporal relationships within text. Par- ticular areas of interest include determining the re- lationship between an event mention and a time ex- pression (timex), as well as determining the relation- ship between two event mentions. The latter, which we refer to as event-event (E-E) temporal classifica- tion is the focus of this work. For a given event pair which consists of two events e1 and e2 found anywhere within an article, we want to be able to determine if e1 happens be- fore e2 (BEFORE), after e2 (AFTER), or within the same time span as e2 (OVERLAP). Consider this sentence1: At least 19 people were killed and 114 people were wounded in Tuesday’s southern Philippines airport blast, officials said, but reports said the death toll could climb to 30. (1) Three event mentions found within the sentence are bolded. We say that there is an OVERLAP rela- tionship between the “killed – wounded” event pair as these two events happened together after the air- port blast. Similarly there is a BEFORE relationship between both the “killed – said”, and “wounded – said” event pairs, as the death and injuries happened before reports from the officials. Being able to infer these temporal relationships allows us to build up a better understanding of the text in question, and can aid several natural lan- guage understanding tasks such as information ex- traction and text summarization. For example, we can build up a …

Content truncated to 500 words