Rules Dependencies in Backward Chaining of Conceptual Graphs Rules
Conceptual Graphs Rules were proposed as an extension of Simple Conceptual Graphs (CGs) to represent knowledge of form “if A then B”, where A and B are simple CGs. Optimizations of the deduction calculus in this KR formalism include a Backward Chaining that unifies at the same time whole subgraphs of a rule, and a Forward Chaining that relies on compiling dependencies between rules.
In this paper, we show that the unification used in the first algorithm is exactly the operation required to compute dependencies in the second one. We also combine the benefits of the two approaches, by using the graph of rules dependencies in a Backward Chaining framework.
KeywordsConceptual Graph Concept Type Graph Homomorphism Partial Projection Conjunctive Type
Unable to display preview. Download preview PDF.
- 4.Baget, J.-F.: Improving the forward chaining algorithm for conceptual graphs rules. In: Proc. of KR2004, pp. 407–414. AAAI Press, Menlo Park (2004)Google Scholar
- 5.Chein, M., Mugnier, M.-L.: Conceptual Graphs: Fundamental Notions. Revue d’Intelligence Artificielle 6(4), 365–406 (1992)Google Scholar
- 10.Salvat, E.: Theorem proving using graph operations in the conceptual graphs formalism. In: Proc. of ECAI 1998, pp. 356–360 (1998)Google Scholar
- 11.Salvat, E., Mugnier, M.-L.: Sound and Complete Forward and Backward Chainings of Graph Rules. In: ICCS 1996. LNCS (LNAI), vol. 1115, pp. 248–262. Springer, Heidelberg (1996)Google Scholar
- 12.Sowa, J.F.: Conceptual Graphs. IBM Journal of Research and Development (1976)Google Scholar