Learning Event Durations from Event Descriptions

Proceedings of the 21st International Conference on Computational Linguistics and 44th Annual Meeting of the ACL, pages 393–400, Sydney, July 2006. c⃝2006 Association for Computational Linguistics Learning Event Durations from Event Descriptions Feng Pan, Rutu Mulkar, and Jerry R. Hobbs Information Sciences Institute (ISI), University of Southern California 4676 Admiralty Way, Marina del Rey, CA 90292, USA {pan, rutu, hobbs}@isi.edu Abstract We have constructed a corpus of news ar- ticles in which events are annotated for estimated bounds on their duration. Here we describe a method for measuring in- ter-annotator agreement for these event duration distributions. We then show that machine learning techniques applied to this data yield coarse-grained event dura- tion information, considerably outper- forming a baseline and approaching hu- man performance. 1 Introduction Consider the sentence from a news article: George W. Bush met with Vladimir Putin in Moscow. How long was the meeting? Our first reaction to this question might be that we have no idea. But in fact we do have an idea. We know the meeting was longer than 10 seconds and less than a year. How much tighter can we get the bounds to be? Most people would say the meet- ing lasted between an hour and three days. There is much temporal information in text that has hitherto been largely unexploited, en- coded in the descriptions of events and relying on our knowledge of the range of usual durations of types of events. This paper describes one part of an exploration into how this information can be captured automatically. Specifically, we have developed annotation guidelines to minimize dis- crepant judgments and annotated 58 articles, comprising 2288 events; we have developed a method for measuring inter-annotator agreement when the judgments are intervals on a scale; and we have shown that machine learning techniques applied to the annotated data considerably out- perform a baseline and approach human per- formance. This research is potentially very important in applications in which the time course of events is to be extracted from news. For example, whether two events overlap or are in sequence often de- pends very much on their durations. If a war started yesterday, we can be pretty sure it is still going on today. If a hurricane started last year, we can be sure it is over by now. The corpus that we have annotated currently contains all the 48 non-Wall-Street-Journal (non- WSJ) news articles (a total of 2132 event in- stances), as well as 10 WSJ articles (156 event instances), from the TimeBank corpus annotated in TimeML (Pustejovky et al., 2003). The non- WSJ articles (mainly political and disaster news) include both print and broadcast news that are from a variety of news sources, such as ABC, AP, and VOA. In the corpus, every event to be annotated was already identified in TimeBank. Annotators were instructed to provide lower and upper bounds on the duration of the event, encompass- ing 80% of the possibilities, excluding anoma- lous cases, and taking the entire context of …

Content truncated to 500 words