Communication support system adapted to a user's background knowledge and the topic context

摘要

This paper proposes a technique for interpolating the information in a real-time conversation via computer between a speaker and listener to compensate for differences in the knowledge presumed for the conversation and introduces Telme, a communications support system that uses this technique. The basic premises are that the system infers the context of the current topic based on the terms used by the speaker and infers the listener's degree of knowledge from the listener's questions and a history of operations such as operations for saving certain data. It then supplements the text information of the words dictated by the speaker with information that the listener should require based on those inferences and displays it on the screen. The degree of knowledge is inferred based on the user's conceptual space, which was generated by using multiple conceptual spaces that were created based on the views and opinions of specialists in various fields. The user's conceptual space, which consists of a set of key words having interrelationships, is adapted to the user by functions for inferring those relationships and the user's degree of knowledge through the user's interactions with the computer. This paper proposes a technique, explains basic mechanisms, and reports on and examines application examples and evaluation experiments.