Creating, testing and implementing a method for retrieving conversational inference with ontological semantics and defaults
Conversational inference refers to that information which is assumed to be understood by both speaker and listener in conversation. With conversational inference, a speaker makes the assumption that what is being omitted from the conversation is already known by the listener. In return, a listener assumes that the information that the listener perceives to be omitted is the same as what the speaker believes to be omitted. Ontological Semantic defaults represent the information which is implied in a single event. Defaults are typically excluded from conversation unless new information is being presented or the speaker is purposefully emphasizing the default for some reason. Little research has been done in the area of defaults. This thesis expands the research on defaults through the implementation and adjustment of an algorithm for default detection. The investigation into default detection is broken into two phases. In the first phase, the original algorithm for default detection is implemented. This algorithm involves pulling defaults based on adjectival modifiers to an object associated with an event. Phase 2 expands the algorithm from Phase 1 to include several additional modifiers. The algorithm from Phase 2 is found to be more effective than that in Phase 1.
Taylor, Purdue University.
Off-Campus Purdue Users:
To access this dissertation, please log in to our