Representing Implicit And Explicit Time Relations In Narrative

This paper describes a representation for time relations in narrative. The time relations are based on both explicit sources of time information (e.g., adverbial expressions or tense) and inplicit sources, such as multiple reference to a single event, narrative time progression and earlier events implied by change of state words. Natural language processing is used to analyse the input text into a set of subject-verb-object units, connected by binary connectives; these units correspond to the events of the narrative. With each event is associated a time in relation to another event, adjusted by an optional time quantity. These time relations have a natural representation as a directed graph whose nodes are time points and whose edges are time intervals. The algorithm for extracting the time relations from a text is illustrated for an excerpt from a hospital discharge summary. I Introduction An understanding of time relations is essential to understanding narrative. For a computer program to "understand" narrative, it must process these time relations; this requires both a representation of the types of time information that interconnect narrated events, and an algorithm to extract this information from the natural language text. The representation of time relations in narrative raises a number of complex problems. On the one hand, there are a multitiude of time words: prepositions (e.g., during, before), adjectives (current, prior), adverbs (then, now), and nouns (duration, week). These various words combine into time expressions (generally adverbial expressions) which must be analysed and reduced to a systematic representation. On the other hand, most events in a narrative do not have an explicit time stated in the form of an adverbial expression. Nonetheless the reader of a narrative can determine the sequence of events even in the absence of explicit time information for each event. To do this, the reader makes use of certain characteristics of narrative discourse, in order to assign time relations to events whose time is not explicitly given. One characteristic of connected discourse is reference: by identifying various mentions of an event as referring to the same event, the reader associates these mentions with the same time. The conventions of narrative discourse also provide certain default time relations between consecutive clauses and/or sentences. For example, the convention of "narrative time progression" provides that in narrative, time does not move backward unless an explicit time marker is provided. An algorithm for capturing time relations in narrative must …