COMMON-SENSE KNOWLEDGE BASED POST-PROCESSING TECHNIQUE IN DATA MININ

摘要

PURPOSE: A common sense-based post-processing method for data mining result relation rules is provided to supply an intellectual data mining result by applying a common sense-based semantic access method to data mining technology. CONSTITUTION: Relation rules generated with a data mining result are defined as a relation vector composed of an antecedent, a consequent, and components based on vector space models. A common sense included in a common sense network of a domain is defined as a common sense vector composed of a pre-concept, a post-concept, and components based on the vector space models. Similarity between the relation rules and the common sense is calculated by calculating a cosine distance between a component vector corresponding to a common sense vector and a relation vector.