Some Reflections on the Task of Content Determination in the Context of Multi-Document Summarization of Evolving Events

摘要

Despite its importance, the task of summarizing evolving events has receivedsmall attention by researchers in the field of multi-document summariztion. Ina previous paper (Afantenos et al. 2007) we have presented a methodology forthe automatic summarization of documents, emitted by multiple sources, whichdescribe the evolution of an event. At the heart of this methodology lies theidentification of similarities and differences between the various documents,in two axes: the synchronic and the diachronic. This is achieved by theintroduction of the notion of Synchronic and Diachronic Relations. Thoserelations connect the messages that are found in the documents, resulting thusin a graph which we call grid. Although the creation of the grid completes theDocument Planning phase of a typical NLG architecture, it can be the case thatthe number of messages contained in a grid is very large, exceeding thus therequired compression rate. In this paper we provide some initial thoughts on aprobabilistic model which can be applied at the Content Determination stage,and which tries to alleviate this problem.