On the Convergence of Incremental Knowledge Base Construction
Ripple Down Rules is a practical methodology to build knowledge-based systems, which has proved successful in a wide range of commercial applications. However, little work has been done on its theoretical foundations. In this paper, we formalise the key features of the method. We present the process of building a correct knowledge base as a discovery scenario involving a user, an expert, and a system. The user provides data for classification. The expert helps the system to build its knowledge base incrementally, using the output of the latter in response to the last datum provided by the user. In case the system’s output is not satisfactory, the expert guides the system to improve its future performance while not affecting its ability to properly classify past data. We examine under which conditions the sequence of knowledge bases constructed by the system eventually converges to a knowledge base that faithfully represents the target classification function. Our results are in accordance with the observed behaviour of real-life systems.
Unable to display preview. Download preview PDF.
- 6.Compton, P., Ramadan, Z., Preston, P., Le-Gia, T., Chellen, V., Mullholland, M.: A trade-off between domain knowledge and problem solving method power. In: Gaines, B., Musen, M. (eds.) 11th Banff KAW Proceeding, pp. 1–19 (1998)Google Scholar
- 10.Kearns, M.J.: Efficient noise-tolerant learning from statistical queries. In: Proceedings of the 25th ACM Symposium on the Theory of Computing, pp. 392–401. ACM Press, New York (1993)Google Scholar
- 12.Motoda, H., Liu, H.: Data reduction: feature aggregation. In: Handbook of data mining and knowledge discovery, pp. 214–218. Oxford University Press, Inc., Oxford (2002)Google Scholar
- 13.Preston, P., Edwards, G., Compton, P.: A 2000 rule expert system without a knowledge engineer. In: Gaines, B., Musen, M. (eds.) 8th Banff KAW Proceeding (1994)Google Scholar
- 15.Scheffer, T.: Algebraic foundation and improved methods of induction of ripple down rules. In: Pacific Rim Workshop on Knowledge Acquisition Proceeding (1996)Google Scholar
- 16.Shiraz, G., Sammut, C.: Combining knowledge acquisition and machine learning to control dynamic systems. In: Proceedings of the 15th International Joint Conference in Artificial Intelligence (IJCAI 1997), pp. 908–913. Morgan Kaufmann, San Francisco (1997)Google Scholar