Metrics of novelty and informativeness of text
Principal Investigator: Julia (Taylor) Rayz
Most communication relays information. Humans have a clear sense of novelty and informativeness in text but there is no usable computational model for this functionality. Our general goal is to design, implement and evaluate a cognitive and computational semantic model for processing information flow in text and determining the extent of novelty, informativeness, focus and saliency of the incoming text relative to the already processed text and stored information from it. Within this project, we explore the constitutive properties of novelty and informativeness in information acquisition and its computational semantic processing. The three specific aims for this research are to: (1) identify the main principles of detecting new information in the incoming text relative to the information already processed from prior texts; (2) identify the principles of calculating the informativeness of the incoming text relative to the already processed information; and (3) determine to what extent novelty and informativeness necessitate changes in the world model.
Other Faculty: Victor Raskin
Students: David Hersh