摘要：The plot of a narrative addresses what happened, and why. While a number of interesting theories of plot have been explored, it has proved hard in narrative interpretation to automatically compute a representation of the plot. This talk describes how to build a representation of what happened by summarizing temporal chains of events that involve a particular protagonist. These chains, which are based on the work of Chambers, can be summarized by various methods, including pruning subgraphs in the representation. Linguistic challenges include habitual expressions and non-literal language. The talk concludes with suggestions for how to layer causal information on top of the representation of what happened.
Loading...
