Bisson, G. (1990) KBG A Knowledge-Based Generalizer, 7th ICML-90, Morgan Kauffmann, 9-15 Borgida, A., Brachman, R.J., McGuinness, D.L., & Resnick, L.A. (1989) Classic: A structural data model for objects, in Proc. ACM SIGMOD-89, Portland, Oregon, 58-67.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....examples describing the behavior of this set of rules. Such behavioral examples are automatically derived from examples of the problem domain ; then inductive learning is used to extract new rules from these behavioral examples (many algorithms perform learning from examples ; see among others [9, 11, 8, 2, 15]) This paper is organized as follows: Section 2 formalizes the transition from examples about the problem domain into examples about the behavior of a given KB over the problem domain. This transition is performed by a redescription operator we call reduction. Section 3 shows how induction from ....

....the R Gamma s standing for the non matching of rule R. These descriptors enable completely new rules to be discovered, such as default rules: Default Class( GammaR Gamma 1 ( R Gamma L ( 2 Space limitations prohibit replicating pseudo code descriptions of induction; see [9, 2, 15] among others. 3.3 Requirements This induction based approach suffers the general machine learning requirements : it needs examples, typical examples and well described examples 3 . So we first need examples of conflicts among rules. Second, our set of descriptors must enable a sufficiently ....

G. Bisson. KBG A Knowledge Based Generalizer. In R. Porter and B. Mooney, editors, 7th International Conference on Machine Learning, pages 9--15. Morgan Kaufman, 1990.

....existing systems such as FOIL[10] FOCL[9] CIGOL[7] Progol[8] need pre specified concepts and pre labeled examples. On the other hand, unsupervised learning systems mainly focus on attribute value concepts. Only a few of them, such as CLUSTER S[13] MOBAL[15] KLUSTER[5] LABYRINTH[14] and KBG[2,3], can learn more expressive concepts. However, they focused on different aspects than ours (a detailed analysis is given in the related work section) The learning task we are interested in can be summarized as follows: Given a relational database along with its schema, a learning system must find ....

....0.0 5) 14 0.4 lll) 13 mm) jjj 0.0 5) 14 0.6 kkk) 14 mm) jjj 0.0 5) 15 0.3 mmm) lll 0.0 7) 12 0.5 mmm) nnn 0.1 4) 15 0.4 nnn) 15 0.6 ooo) 16 0.5 ppp) 16 0. 7 ppp) Its schema and value ranges: T1: C11 char(2) C12 integer[2 7] C13 float[0.4 0.8] T2: C21 integer[12 17] C22 float[0.1 0.7] C23 char(3) T3: C31 integer[13 16] C32 char(2) T4: C41 char(3) C42 float[0.0 0.1] C43 integer[4 7] Figure 2: An abstract database To illustrate the idea, consider for ....

Bisson, G., "KBG, A Knowledge Based Generalizer", The Proceedings of the 7th International Conference on Machine Learning, Morgan Kaufmann, pp. 9-15, 1990.

....relationships nearly always request a strong background theory or the thorough support of the expert [3] In the field of analogy, the structural mapping of Gentner [10] proposes an evaluation of the degree of analogy between two cases, based on a cognitive approach. In inductive learning, Bisson [6] developed a similarity measure in order to cluster and classify first order examples. A commonly shared opinion is that much knowledge is hidden in a (good) similarity measure. Reversing this claim, we propose to compile a knowledge base into a similarity measure in a 3 step process : first, a ....

....similarity measure thus only requires to use a redundant learner. Remark. Note that this approach applies whatever the initial formalization of the domain : it only needs this formalism to be tractable for a redundant learner. This requirement holds for propositional [9, 24] and first order logic [6, 25]. 2.4 Accounting for Adaptation The rule based similarity can use humanly provided rules as well as rules learned from a data set ; it can also combine both. However, the available knowledge if any is usually not sufficient to build a usable similarity. One then has to extract a rule set from the ....

G. Bisson, KBG A Knowledge Based Generalizer. 7

No context found.

Bisson, G. (1990) KBG A Knowledge-Based Generalizer, 7th ICML-90, Morgan Kauffmann, 9-15 Borgida, A., Brachman, R.J., McGuinness, D.L., & Resnick, L.A. (1989) Classic: A structural data model for objects, in Proc. ACM SIGMOD-89, Portland, Oregon, 58-67.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC