A Declarative Similarity Framework for Knowledge Intensive CBR by Díaz-Agudo and González-Calero Presented by Ida Sofie G Stenerud 25.October 2006
Terminology CBROnto Description Logic LOOM
CBROnto ”A task based ontology compromising common CBR terminology” A vocabulary for expressing the CBR elements CBROnto’s 2 purposes: To integrate CBR process knowledge and domain knowledge To be a domain-indepentent framework for designing CBR systems
Description Logics A knowledge representation language Formal objects: Concepts Relations Individuals Reasoning mechanisms Subsumption Instance Recognition
LOOM A description logic implementation A query language for CBR Example:
The Similarity Framework Several similarity measures can coexist at one time Several approaches to case retrieval: Relevance Criteria User-defined or pre-defined in program Similarity Criteria Declarative represetation of differences and similarity Representational approach ”Semantic traversal” of the hierarchy Computational approach Can be a combination of all the above
The Similarity Framework
User-defined Relevance Criteria Relevance: Why is this case more relevant than other cases? More-on-point, Most-on-point
Similarity Terms Explicit computation of similarity terms Declaratively Similarity: ”the most specific concept which subsumes 2 cases”
Representational Approach Assignment of similarity meanings to the path between cases The Generic Travel Operator
Computational Approach Different alternatives to compute numeric similarity between attributes Nearest Neighbour Algorithm The classic global similarity approach For common attributes: use local measure For common relations: use global measure to compare related sub-objects Similarity is a weighted sum of these NB: Only attribute level, not instance level similarity!
Intra-class vs. inter-class similarity Intra-class similarity: Dependent on the attribute fillers Inter-class similarity Dependent on object position in hierarchy Multiplied to get final similarity result
The Similarity Framework
Similarity Functions Local similarity Similarity between two values of a type Global similarity Defines how you combine local similarities Positional similarity Indepentent of attribute values, only dependent on position in hierarchy
A Case Matching Example Case 2 Color: Dark Gray Price: Case 1 Color: Dark Green Price:
Discussion Any questions?