A knowledge representation for conceptual information retrieval

Publisher Website
Google Scholar
Abstract
This article contains a detailed description of the knowledge representation used in RUBRIC, our concept-based, full-text information retrieval system. Since we view information retrieval as a process of evidential reasoning, the RUBRIC knowledge representation is designed to help maintain a clear distinction between descriptions of domain concepts, the documentary evidence for those concepts, and the beliefs about concepts that can be sustained on the basis of the evidence. This article is a consolidation of our earlier work in this area, containing, as it does, a revised language for defining concepts, an extended view of the role of evidence, and a formally grounded model of belief.

This publication has 5 references indexed in Scilit: