Authors:
Steven B. Kraines
and
Weisen Guo
Affiliation:
University of Tokyo, Japan
Keyword(s):
Knowledge Representation, Semantic Matching, Semantic Similarity, Logic Inference.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Communication, Collaboration and Information Sharing
;
Innovation Facilitation
;
Intelligent Information Systems
;
KM Strategies and Implementations
;
Knowledge Management and Information Sharing
;
Knowledge-Based Systems
;
Metadata and Structured Documents
;
Social Networks and the Psychological Dimension
;
Studies, Metrics & Benchmarks
;
Symbolic Systems
;
Tools and Technology for Knowledge Management
Abstract:
If researchers created computer-understandable descriptors as part of the process of authoring journal articles and other expert knowledge resources, intelligent computer-aided matching and searching applications that are critical for addressing complex and large-scale problems in society could be realized. The EKOSS system enables knowledge experts to create computer-understandable descriptors of their knowledge resources using description logics ontologies as formal knowledge representation languages. The descriptors, called semantic statements, are authored as description logic ABoxes in reference to a shared domain ontology in the form of a TBox. Reasoners using logic-based inference can then measure the semantic similarity between semantic statements, which can be applied in knowledge searching, mining and integration applications. A method for semantic matching that uses logic inference based on a DL ontology TBox to increase both the precision and recall of matching descriptor
s created as ABoxes is described, and the accuracy of the method compared to matching without logic inference is analyzed between a set of 15 semantic statements created using EKOSS to describe research articles related to sustainability science.
(More)