CaseBased Solution Diversity Alexandra Coman Hctor MuozAvila Dept
Case-Based Solution Diversity Alexandra Coman Héctor Muñoz-Avila Dept. of Computer Science & Engineering Lehigh University Sources: • cbrwiki. fdi. ucm. es/ • www. iiia. csic. es/People/enric/AICom. html • www. cse. lehigh. edu/~munoz/CSE 335/ • www. aic. nrl. navy. mil/~aha/slides/ • http: //www. csi. ucd. ie/users/barry-smyth • http: //www. csi. ucd. ie/users/lorraine-mcginty
Outline n n n Lehigh University The In. Sy. Te Laboratory Overview of Case-Based Reasoning q q q n n Similarity Retrieval Adaptation Conversational Case-based reasoning Diversity versus Similarity General versus Episodic Knowledge Final Remarks
Synthetizing Diversity n n Showcasing diverse solutions: success story in recommender systems (Smyth, Burke, Mc. Ginty …) Plan diversity: q q n Synthetizing diversity: q 11 Definition of the problem: quantitative vs qualitiative (Myers, AAAI-01) Generating two or more quantitative different plans for same problem (Srivastava et al, IJCAI-07) q q Case-based retrieval and adaptation from plan library (Coman & Munoz-Avila, ICCBR-10; 11 – under review ) Generating two or more qualitatively different plans for same problem (Coman & Munoz-Avila, AAAI-11) Our common solution: n S: diverse solutions so far, s: candidate solution, P: new problem sim(s, P) + relative. Diversity(s, S) n What changes: S, s, P, sim(), D(s, s’)
Research Program: Synthetizing Diversity preliminary work: New insight: Proposed idea: Research topics: Plan Diversity Case-based plan diversity sim(s, P) + relative. Diversity(s, S) + cost(s) • Representation scope of using D() versus qualitative diversity • Trade-offs of solution: • Diversity versus quality • Diversity versus generation • Diversity in other paradigms: search (A*) Danger: don’t want it to be a planning proposal
Focus Point: Diversity in CBR
Traditional Retrieval Approach § Similarity-Based Retrieval § Select the k most similar items to the current query. Query Available case Similar case § Problem § Vague queries. § Limited coverage of search space in every cycle of the dialogue. Lorraine Mc. Ginty and Barry Smyth Department of Computer Science, University College Dublin C 2 C 3 C 1 Q
Diversity Enhancement § Diversity-Enhanced Retrieval § Select k items such that they are both similar to the current query but different from each other. Query Available case Retrieved case § Providing a wider choice allows for broader coverage of the product space. § Allows many less relevant items to be eliminated. Lorraine Mc. Ginty and Barry Smyth Department of Computer Science, University College Dublin C 1 Q C 2 C 3
§ Dangers of Diversity Enhancement Leap-Frogging the Target § Problems occur when the target product is rejected as a retrieval candidate on diversity grounds. § Protracted dialogs. § Diversity is problematic in the region of the target product. § Use similarity for fine-grained search. § Similarity is problematic when far from the target product. § Use diversity to speed-up the search. Lorraine Mc. Ginty and Barry Smyth Department of Computer Science, University College Dublin C 1 Q C 2 T C 3
Final Remarks
Questions?
- Slides: 10