Modeling Context in Scenario Template Creation

Abstract

We describe a graph-based approach to Scenario Template Creation, which is the task of creating a representation of multiple related events, such as reports of different hurricane incidents. We argue that context is valuable to identify important, semantically similar text spans from which template slots could be generalized. To leverage context, we represent the input as a set of graphs where predicate-argument tuples are vertices and their contextual relations are edges. A context-sensitive clustering framework is then applied to obtain meaningful tuple clusters by examining their intrinsic and extrinsic similarities. The clustering framework uses Expectation Maximization to guide the clustering process. Experiments show that: 1) our approach generates high quality clusters, and 2) information extracted from the clusters is adequate to build high coverage templates. 1