WHAT ONLINE MACHINE LEARNING CAN DO FOR KNOWLEDGE ACQUISITION - A CASE-STUDY

This paper reports on the development of a realistic knowledge-based a pplication using the MOBAL system. Some problems and requirements resu lting from industrial-caliber tasks are formulated. A step-by-step acc ount of the construction of a knowledge base for such a task demonstra tes how the interleaved use of several learning algorithms in concert with an inference engine and a graphical interface can fulfill those r equirements. Design, analysis, revision, refinement and extension of a working model are combined in one incremental process. This illustrat es the balanced cooperative modelling approach. The case study is take n from the telecommunications domain and more precisely deals with sec urity management in telecommunications networks. MOBAL would be used a s part of a security management tool for acquiring, validating and ref ining a security policy. The modeling approach is compared with other approaches, such as KADS and stand-alone machine learning.