Knowledge acquisition is usually the first step in building ontologies. On the one hand, knowledge is typically implicitly contained in large collections of unstructured documents. Therefore it is extremely troublesome to manually identify relevant concepts. On the other hand, users are often not fully satisfied with the results of automated stateof-the-art ontology learning techniques. In this paper we present a technique for large-scale Knowledge Acquisition supported Semi-automated Ontology building (KASO) and a corresponding software system. By applying KASO and using this software, users are able to bootstrap the process of building high quality ontologies by automatically acquiring concepts from large-scale document collections and to make use of traditional knowledge acquisition approaches to refine and organize the machine-generated concepts. Evaluation studies and user experiences indicate the applicability of KASO in bootstrapping ontology construction.
The acquisition of knowledge about a task can be viewed as a process of incorporating new knowledge into some existing knowledge structure [Rosenbloom 1988]. The existing knowledge can guide and constrain the search for new knowledge, and the process of integrating the new knowledge with the old may identify additional opportunities for learning. An acquisition system that takes this view needs to represent and understand the knowledge about the task as well as the process of finding and integrating new knowledge. A general trend in research on knowledge acquisition has been to make the knowledge structures that guide acquisition increasingly explicit. In early acquisition tools many of the requirements that needed to be satisfied when adding a new piece of knowledge were not stated.
This paper describes a framework, currently under development, for using Software Visualization (SV) technology to drive the knowledge acquisition process. Software visualization (Price et al. 1993) is the use filmcraft, cartoon animation and graphic design techniques to display data structures, programs, and algorithms. It has successfully been used to validate Knowledge Based Systems (KBS)(Domingue 1995), aid expert (Eisenstadt & Brayshaw 1988) and novice programmers (Lieberman & Fry 1995) (Mulholland"1995), and to teach computer science students about algorithms (Brown 1988). The framework proposed here is based around an Expert Scripted Visualization (ESV) which experts can create in domain dependent visualization scripting environment. The scripting environment allows experts to set up and solve test cases by direct manipulation.
Expert system projects are often based on collaboration with single domain expert. This leads to difficulties in judging the suitability of the chosen task and in acquiring the detailed knowledge required to carry out the task. This anecdotal article considers some of the advantages of using a diverse collection of domain experts.