S. Craw and D. Sleeman. Knowledge-based refinement of knowledge based systems. Technical Report 95/2, The Robert Gordon University, Aberdeen, UK, 1995.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....expert will get insight into the situations where his expectations differ from the actual behaviour of the system. In such cases, revision is necessary. Revision may be carried out manually or it could also be carried out automatically using knowledge revision systems (see e.g. Craw s Krust system [5], or De Raedt s Clint [6] After revision, the validation and verification process can be repeated until the human expert is satisfied with the results of the induction engines. 4 Multi agent systems The agent oriented approach to programming as used in the RoboCup environment brings a number of ....

....This work builds upon earlier ideas on combining verification and validation with inductive logic programming [8] It is also related to other approaches applying machine learning with validation and verification. This includes the work of Susan Craw on her Krust system for knowledge refinement [5], the work by Bergadano et al. and the work by De Raedt et.al. 11] The approach taken in Krust is complementary to ours. Rather than starting from examples of the actual behaviour of the system, Krust starts from examples of the desired behaviour of the system. Whenever the two behaviours do not ....

S. Craw and D. Sleeman. Knowledge-based refinement of knowledge based systems. Technical Report 95/2, The Robert Gordon University, Aberdeen, UK, 1995.

....expert will get insight into the situations where his expectations differ from the actual behaviour of the system. In such cases, revision is necessary. Revision may be carried out mannualy or it could also be carried out automatically using knowledge revision systems (see e.g. Craw s Krust system [4], or De Raedt s Clint [5] After revision, the validation and verification process can be repeated and this until the human expert is satisfied by the results of the induction engines. 4 Multi agent systems 4.1 Complexity of Multi Agent Systems Recently a lot of attention has been devoted to ....

....much. 6 Related work This work buils upon earlier ideas on combining V V with inductive logic programming [7] It is also related to other approaches applying machine learning with validation and verification. This includes the work of Susan Craw on her KRUST system for knowledge refinement [4], the work by Bergadano et al. and the work by De Raedt et al. 10] The approach taken in KRUST is complementary to ours. Rather than starting from examples of the actual behaviour of the system, KRUST starts from examples of the desired behaviour of the system. Whenever the two behaviours do not ....

S. Craw and D. Sleeman. Knowledge-based refinement of knowledge based systems. Technical Report 95/2, The Robert Gordon University, Aberdeen, UK, 1995.

....expert will get insight into the situations where his expectations differ from the actual behavior of the system. In such cases, revision is necessary. Revision may be carried out mannualy or it could also be carried out automatically using knowledge revision systems (see e.g. Craw s Krust system [6], or De Raedt s Clint [7] After revision, the validation and verification process can be repeated until the human expert is satisfied with the results of the induction engines. Very simple verification tasks can be handled by propositional systems (e.g. C4.5 [15] or CN2 [4] However, in most ....

....much. 6 Related work This work buils upon earlier ideas on combining V V with inductive logic programming [9] It is also related to other approaches applying machine learning with validation and verification. This includes the work of Susan Craw on her KRUST system for knowledge refinement [6], the work by Bergadano et al. and the work by De Raedt et al. 12] The approach taken in KRUST is complementary to ours. Rather than starting from examples of the actual behavior of the system, KRUST starts from examples of the desired behavior of the system. Whenever the two behaviors do not ....

S. Craw and D. Sleeman. Knowledge-based refinement of knowledge based systems. Technical Report 95/2, The Robert Gordon University, Aberdeen, UK, 1995.

....(TR) as described by Meseguer [16] Theory refinement considers the improvement of an approximate domain theory from a set of cases with known solution. Adopting this definition of TR it can be seen as the basis for KB debugging or updates as well as for KB evolution in a slowly changing domain [7]. Since according to Craw Sleeman [8] such KB maintenance activities can be understood as refinement in response to validation and thus performed by collaboration of exploration and validation algorithms, analysis of existing TR systems is a good starting point for our research concerning ....

....goal of approaching a methodology for KB maintenance, we can check techniques and methods already existing in TR against requirements coming from KBS development and use. This leads to suggestions for further work in TR opening a wider range of application for TR technology. As Craw and Sleeman [7] mentioned, the classical concept of TR stemming from data intensive concept learning (cf. 19] is not adequate for the refinement of knowledge based systems. Here, we rather have a data sparse process exploiting various sources of background knowledge in order to manage continuous change in ....

[Article contains additional citation context not shown here]

S. Craw and D. Sleeman, `Knowledge-based refinement of knowledge based systems', Technical Report TR9505, University of Aberdeen, Department of Computing Science, (1995).

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