Automatic Event Reference Identification

Automatic Event Reference Identification Olivia March CSSE University of Melbourne Victoria 3010 Australia oliviacm@csse.unimelb.edu.au Timothy Baldwin NICTA Victoria Research Laboratories University of Melbourne Victoria 3010 Australia tim@csse.unimelb.edu.au Abstract Event reference identification is often treated as a sentence level classification task. However, several different event ref- erences can occur within a single sentence. We present a set of experiments involving real world event reference identification at the word level in newspaper and newswire documents, addressing the issue of effec- tive text representation for classification of events using support vector machines. Our final system achieved an F-score of 0.764, significantly exceeding that of our base- line system. Additionally we achieved a marginally higher performance than a more complex comparable system. 1 Introduction Multi-document summarization seeks to com- bine the salient information from multiple inde- pendent documents into a single coherent sum- mary. The Document Understanding Conference (DUC) shared tasks are a measured attempt to ap- ply multi-document summarization to newswire document sets, to produce a fluent 250 word sum- mary (Dang, 2006). The resultant summaries are then scored and assessed, depending on the specifics of the task. Newswire and newspaper document sets such as those used in DUC often describe the same series of events across multiple sources, for ex- ample, media coverage of an election in Bolivia. Here, a time-line can be a more effective way of structuring a multi-document summary than a traditional paragraph-style summary. More gen- erally, historical and biographical information is frequently presented as a time-line of events, and the ability to produce such time-lines from in- dependent sources is valuable for online content providers such as Wikipedia. The following is an illustration of the utility of time-lines in analysing topics (events highlighted in bold to illustrate their importance in time-line generation): The Russian Revolution in 1917 was triggered by a combination of eco- nomic breakdown, war weariness, and discontent with the autocratic sys- tem of government, and it first brought a coalition of liberals and moderate so- cialists to power, but their failed poli- cies led to seizure of power by the Communist Bolsheviks on October 25. Existing summarization systems, such as the DUC 07 main task system of Stokes et al. (2007), are effective at general summary tasks but have difficulty with event based summaries. For exam- ple, Stokes et al. (2007) performs well on general questions centred on a specified topic such as: What attacks have occurred against tourists in Luxor, Egypt? but not so well on topics requiring a listing of events, such as: What have been the most recent sig- nificant events in the life and career of actress Angelina Jolie? In this paper, we identify the most effective rep- resentation of text for real-world event reference identification in newspaper and newswire text us- ing support vector machines (SVMs). The work presented here is the basis of an event-based sum- mary system which will in turn form a component of the Stokes et al. (2007) system, where each question will be classified as …

Content truncated to 500 words