Par-KAP: a Knowledge Acquisition Tool for Building Practical Planning Systems

Recently, attention has been focused on provid­ ing Knowledge Acquisition (KA) support for building practical planning systems. Such sup­ port is needed to guide a knowledge engineer in selecting planning methods, as well as for build­ ing and validating the planning knowledge-base for a given practical domain. Following cur­ rent practice in knowledge acquisition, devel­ oping KA tools for planning requires that a number of planning knowledge components are made explicit. This includes explicating (i) a planning domain ontology, (ii) a library of problem-solving methods (PSMs) used in plan­ ning, and (iii) a set of domain requirements that are used to select a suitable PSM. In this paper, we summarize the planning knowledge components which we have identified in previ­ ous work, and, based on these, present an im­ plementation (Par-KAP) that can exploit these models to aid knowledge engineers in construct­ ing practical planning systems. 1 Introduction Constructing a planner for a particular application is a difficult job, for which few knowledge acquisition tools currently exist. Due to an increasing need for the build­ ing of planning systems that can handle real world ap­ plications, knowledge engineering efforts need to focus on the following questions: 1. Is a particular practical planning system from the literature suitable for a given domain? 2. What type of domain knowledge is required for the application and how can it be represented? 3. What are the most suitable planning methods for the domain? 4. What control strategy can best satisfy a desired sys­ tem performance? Although much is known about planning systems, and the literature is extensive (cf. [Tate et a/., 1990; Tate, 1996], the focus of past work has primarily been on the development of planning systems, as opposed to an­ swering questions such as those above. At the same time, while knowledge acquisition research has focused on such questions for other Al tasks [Schreiber et al., 1994], lit­ tle attention has been given to planning. In is only re­ cently that research has started to focus on this sort of This work primarily builds on modern approaches to Knowledge Acquisition which stress the importance of libraries with reusable modeling components to support the knowledge engineer in constructing the required system model [Breuker and van de Velde, 1994]. Examples of library constituents in­ clude domain models, domain ontologies, generic tasks, problem-solving methods, inference structures, control knowledge, etc. This paper builds on the work …