R. Bergmann and W. Wilke. Building and Refining Abstract Planning Cases by Change of Representation Language, Journal of Artificial Intelligence Research (JAIR), 1995, 3:53-118.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....generation process is automatic. Our system does not implement case deletion policy; this will be addressed in the future research. The summary of the work presented in this paper with respect to the framework by Bergman and Wilke is shown in Table 6. Similar to the approach in the PARIS system [9], the recognizer does not form abstraction by simply dropping sentences. However, the simplicity of our abstraction scheme allows for the minimal abstraction theory, since we are simply counting a number of occurrences of a literal of a certain type among a collection of literals representing a ....

Bergmann, R., & Wilke, W. (1995). Building and Refining Abstract Planning Cases by Change of Representation Language. Journal of Artificial Intelligence Research, 3:53---118.

....a complete domain theory. CHEF [Hammond, 1989] and DIAL [Leake et al. 1997] are case based, but do not have a generative component, and thus need a large case base to perform well across a wide variety of problems. Prodigy Analogy [Veloso, 1994] DerSNLP [Ihrig Kambhampati, 1994] and Paris [Bergmann Wilke, 1995] integrate generative and casebased planning, but require a complete domain theory and are not mixed initiative. SIPE II [Wilkins, 1998] is a mixed initiative generative planner, but does not use cases. NaCoDAE [Muoz Avila et al. 1999] is a mixedinitiative case based planner, but does not employ ....

Bergmann R., & Wilke W. Building and refining abstract planning cases by change of representation language. Journal of Artificial Intelligence Research, 3, 53-118, 1995.

....not interact with one another, or should only interact in limited ways. Fortunately, many domains and applications do benefit from decomposable solution structures, as is evident from the popularity of hierarchical problem solving and decompositional design techniques in general (see, for example, [2], 4] 6] 14] 16] 17] 36] 37] 8 RELATED WORK Hierarchical case based reasoning is one way of supporting multiple case reuse. Its distinguishing features include: 1) Using cases at multiple levels of abstraction to represent complex problem solutions as hierarchies; 2) Using abstract ....

R. Bergmann and W. Wilke, "Building and Refining Abstract Planning Cases by Change of Representation Language," J. Artificial Intelligence Research, vol. 3, pp. 53-118, 1995.

....Marefat et al. Marefat and Britanik, 1996] Process Planning MEDIC [ Turner, 1988 ] Diagnosis Therapy MPA [ Ram and Francis Jr. 1996] Navigation Planning MRL [ Koehler, 1994; Koehler, 1995; Koehler, 1996 ] Software Reuse Murdock et al. Murdock et al. 1997 ] Other PARIS [ Bergmann and Wilke, 1995; Bergmann and Wilke, 1996; Veloso et al. 1996 ] Process Planning PLEXUS [ Alterman, 1986; Alterman, 1988] Other PPR [ Plaza et al. 1997] Other PRIAR [ Kambhampati and Hendler, 1992 ] Other PRODIGY ANALOGY [ Veloso, 1994; Veloso et al. 1995; Veloso et al. 1996 ] Logistics ROBBIE [ Fox and ....

....system automatically ranks all the retrieved cases. When a case based planner uses a retrieval method which does not rank the plans, a separate selection task is needed. For example ASP II and TOLTEC use a similarity metric to rank a set of cases retrieved in a previous matching phase. In PARIS [ Bergmann and Wilke, 1995 ] the level of the case in an abstraction hierarchy is used for the selection, the lower the level is, the less the adaptation is. Mu noz Avila et al. 1999a; Mu noz Avila et al. 1999b ] propose an interactive approach, the user guides the selection by answering to a set of questions stored ....

R. Bergmann and W. Wilke. Building and Refining Abstract Planning Cases by Change of Representation Language. Journal of Artificial Intelligence Research, 3:53--118, 1995.

....a complete domain theory. CHEF [Hammond, 1989] and DIAL [Leake et al. 1997] are case based, but do not have a generative component, and thus need a large case base to perform well across a wide variety of problems. Prodigy Analogy [Veloso, 1994] DerSNLP [Ihrig Kambhampati, 1994] and Paris [Bergmann Wilke, 1995] integrate generative and casebased planning, but require a complete domain theory and are not mixed initiative. SIPE II [Wilkins, 1998] is a mixedinitiative generative planner, but does not use cases. NaCoDAE [Muoz Avila et al. 1999] is a mixed initiative case based planner, but does not employ ....

Bergmann R., & Wilke W. Building and refining abstract planning cases by change of representation language. Journal of Artificial Intelligence Research, 3, 53-118, 1995.

....of hierarchical problem solving techniques results in a lower time and space complexity of this architecture with respect to the system it is based on. However, though the hierarchical domain representation model and planning can reduce the benefit of using these techniques in some cases [2, 3, 4, 10], it has clear advantages from the point of view of computer aided design of control programs. On the one hand, this approach provides an easy entry level for end users (control engineers) The hierarchy of agents of a domain accepts SP88 standard descriptions, usually handled by control ....

R. Bergman and W. Wilke, `Building and refining abstract planning cases by change of representation language', JAIR, 3, 53--118, (1995).

....from one or more cases [13] A number of case based reasoning systems have modeled an important aspect of this flexibility: use of case abstractions to guide case indexing, matching, and adaptation. Termed stratified case based reasoning (SCBR) in [4] this approach has been used for planning [3], 12] design of control software [16] and route planning [4] 5] A comparative analysis of various approaches to using abstraction in CBR is set forth in [2] Systematic empirical analyses set forth in [4] and [5] compared the performance of heuristic search (A ) Refinement [11] i.e. ....

....method that is incapable of producing solutions as good as those found through ab initio problem solving. 3 Logistics Planning The utility of using case abstractions for indexing and adaptation has been demonstrated in planning domains characterized by task decomposition abstraction hierarchies [3], 12] However, development of task decomposition hierarchies is, in general, a difficult and time consuming task. The second experiment was designed to explore the feasibility of using SCBR as an indexing technique using inexpensive off the shelf abstraction hierarchies. In the early 1990s, ....

R. Bergmann and W. Wilke, Building and Refining Abstract Planning Cases by Change of Representation Language, Journal of Artificial Intelligence Research, 3:53-118, 1996.

....vocabulary. This vocabulary extends the primitive solution operator set to include abstract operators, which are then used in the abstract cases. In this way D j Vu s case hierarchies share much in common with the type of plan hierarchies that are automatically built by the PARIS system (Bergmann Wilke, 1995, 1996) In particular, both systems make use of abstract knowledge that is represented as actual highlevel solution code. The EADOCS system (Netten Vingerhoeds) also uses abstraction knowledge, however, unlike D j Vu and PARIS this knowledge is not stored with the structure of abstract cases. ....

....of D j Vu s decomposition technique is that it does not relay on static decomposition knowledge. Instead, the case hierarchies drive the decomposition process, and decomposition is a direct result of the retrieval and adaptation of abstract cases. In this context D j Vu is similar to PARIS (Bergmann Wilke, 1995, 1996) because it permits the explicit reuse of previous problem decompositions (the abstract case solutions) and it is these adapted decompositions that guide the decomposition process. However, an important difference between D j Vu and PARIS is that D j Vu refines its abstract solutions by ....

Bergmann, R., & Wilke, W. (1995) Building and Refining Abstract Planning Cases by Change of Representation Language. Journal of Artificial Intelligence Research, 3, pp. 53 - 118.

....relevant cases is ensured without negatively affecting the performance of the case based planning process. 1 Introduction Reusing previous problem solving experience has proven to speed up planning systems. Problem solving experience can be stored in generalized (Minton, 1988) or abstracted (Bergmann Wilke, 1995) form, or it can be stored as cases (Veloso, 1994; Ihrig Kambhampati, 1994; Yang Lu, 1994) One of the methods better studied and used for adapting cases is derivational analogy (Carbonell, 1983; Veloso Carbonell, 1993; Cunningham Slattery, 1994) that basically consists of replaying the ....

Bergmann, R. (1995). Building and refining abstract planning cases by change of representation language. To appear in Journal of AI Research.

....that a previous solution can be usefully applied to a new problem. 1 Stratified Case Based Reasoning Stratified Case Based Reasoning is a technique under which case abstractions are used to assist case indexing, matching, and adaptation. This approach has been applied to case based planning [BW95, KH92], design of control software [SK94, SC92] and route planning [BA95] See [BW96] for a comparative analysis of previous approaches) Use of case abstractions has the following potential benefits: Indexing and retrieval. A more abstract solution to a problem can provide an accurate index to ....

R. Bergmann and W. Wilke. Building and refining abstract planning cases by change of representation language. Journal of Artificial Intelligence Research, 3(53--118), 1995.

....adaptation. Table 1 compares the case based planners CHEF (Hammond, 1986) Priar (Kambhampati Hendler, 1992) CaPER (Kettler et al., 1994) Prodigy Analogy (Veloso, 1994) SPA (Hanks Weld, 1995) MPA (Francis Ram, 1995) derSNLP EBL (Ihrig Kambhampati, 1996) MRL (Koehler, 1994) Paris (Bergmann Wilke, 1995) and CAPlan CbC (Mu noz, 1998) according to the following criteria: ffl Adaptation Method: Many plan adaptation approaches have been investigated, ranging from domain specific heuristic methods to specialized mechanisms that reason about the underlying planning models. ffl Similarity Metrics: ....

Bergmann, R., & Wilke, W. (1995). Building and refining abstract planning cases by change of representation language. Journal of Artificial Intelligence Research, 3, 53--118.

....specifications, which will be solved by the retrieval and adaptation of more detailed cases. In this way the solutions of D j Vu s abstract cases are similar to the abstract solution plans often used by the planning community. This form of operator abstraction is also used by Bergmann Wilke [2]. 4.3 Case Hierarchies 10 We have already introduced the idea that complex problems are to be represented as hierarchical collections of abstract and design cases. It is important at this stage to emphasise that these hierarchies are not automatically created by D j Vu. Instead, they must be ....

....Instead, they must be engineered from complex cases, and so this form of hierarchical casebased reasoning increases the knowledge engineering cost of CBR. However, there is promising research available that leads us to believe that it may be possible to automatically generate these hierarchies ([2], 6] 12] Furthermore, as we will discuss in section 8, there are also distinct advantages associated with HCBR, including a reduced need for adaptation knowledge, and corresponding decrease in the knowledge engineering effort associated with the compilation of this knowledge. Unload 1 ....

R. Bergmann and W. Wilke, "Building and Refining Abstract Planning Cases by Change of Representation Language," Journal of Artificial Intelligence Research, Vol. 3, pp. 53-118, 1995.

.... the performance of planning with OE consistent permutations and of planning with non OE consistent permutations of the goals 9 Related Work Process planning has been the subject of several studies in AI (Hayes, 1987; Kambhampati et al. 1991; Gil, 1991; Karinthi et al. 1992; Nau et al. 1995; Bergmann and Wilke, 1995), involving different AI techniques and different problem solving criteria. Hayes, 1987) already observed that the domain of process planning is characterized by the high number of interactions and uses a domain dependent mechanism to detect interactions before they occur. This idea is similar to ....

....all possible solution plans must be computed for each part of the workpiece, which can be very expensive. In general, merging subplans is NP complete (Yang et al. 1992) Our experiments supports this result as the execution time increased exponentially with the number of subplans being merged. In (Bergmann and Wilke, 1995), a case based planner guiding a linear, state space planner is presented. This system provides levels of abstraction and is developed with the idea that plans differing in details have the same abstract representation. In addition, a specification for the domain of process planning is given. ....

Bergmann, R. and Wilke, W. (1995). Building and refining abstract planning cases by change of representation language. Journal of Artificial Intelligence Research, 3:53-- 118.

....simple chunk substitution is safe with respect to the above named dependencies and to perform it as much as boundary friction will allow. 3. ReVerb 3. 1 Overall Architecture The overall CBR model adopted for this work is best understood by consideration of Figure 2 following (adapted from Bergmann, 1995). While providing a coherent organisation for all the necessary components of a full EBMT system, we presently concentrate on only three components of this model, namely: Case Organisation, Case Retrieval, and Case Adaptation (described in sections 3, 4 and 5 respectively) The central philosophy ....

Bergmann R., Wilke W., (1995) Building and Refining Abstract Planning Cases by change of Representation Language, Journal of AI Research, 3, 53-118.

....approaches retrieve, reuse, and retain cases in a representation at a single level of abstraction. In this predefined representation, cases, new problems, as well as general knowledge must be represented. Recently, some researchers have started to investigate the use of abstraction in CBR (e.g. [19, 12, 3, 4, 5, 20, 2], cf. also [13] p. 576) However, a clear picture of how CBR can benefit from abstraction has not be drawn till now. In AI, the use of abstraction was originally inspired by human problem solving (cf. 14] and has already been successfully used in different fields such as theorem proving, ....

....properties. In traditional hierarchical problem solving (e.g. Abstrips [17] abstraction levels are constructed by simply dropping certain features of the more concrete representation levels. However, it has been shown that this view of abstraction is too restrictive and representation dependent [4, 11] to make full use of the abstraction idea. In general, different levels of abstraction require different representation languages, one for each level. Abstract properties can then be expressed in completely different terms than concrete properties [2] Different Kinds of Cases Based on the level ....

[Article contains additional citation context not shown here]

R. Bergmann and W. Wilke. Building and refining abstract planning cases by change of representation language. Journal of Artificial Intelligence Research, 3:53--118, 1995.

....details are added. Hierarchical adaptation can either reuse a single case or it can reuse different cases for different levels of abstraction or for refining different details of the solution. Examples of systems using hierarchical adaptation are D eJ a V u [Smyth and Cunningham, 1992] Paris [Bergmann and Wilke, 1995], or the stratified CBR approach by [Branting and Aha, 1995] 4 Discussion Adaptation is still a big search issue in case based reasoning that is also getting increasingly important in practice as the complexity of applications increases. During the last years, there have been several workshops ....

Bergmann, R. and Wilke, W. (1995). Building and refining abstract planning cases by change of representation language. Journal of Artificial Intelligence Research, 3:53--118.

....Bergmann and Wolfgang Wilke Centre for Learning Systems and Applications (LSA) University of Kaiserslautern Dept. of Computer Science P.O. Box 3049, D 67653 Kaiserslautern, Germany e mail: bergmann informatik.uni kl.de 1 Introducion Paris (Plan Abstraction and Refinement in an Integrated System) [4, 2] is a domain independent case based planning system which allows the flexible reuse of planning cases by abstraction and refinement. This approach is mainly inspired by the observation that reuse of plans must not be restricted to a single description level. In domains with a high variation in the ....

....is abstracted to a single abstract operator. Note that the above explained kind of case abstraction is performed by an automatic procedure in Paris as part of the retain phase of the CBR cycle. Abstract cases are then stored in the case base, indexed by an abstraction hierarchy (for details see [4]) 2.1 Retrieval and Reuse of Abstract Cases When a new problem must be solved, an abstract case is retrieved which exactly matches the current problem at the abstract level. If several matching abstract cases are contained in the case base, the retrieval procedure (using an abstraction ....

[Article contains additional citation context not shown here]

R. Bergmann and W. Wilke, `Building and refining abstract planning cases by change of representation language', Journal of Artificial Intelligence Research, 3, 53--118, (1995).

....the workpiece from Ascend.2 (arc labeled P) which is the right choice. u cut1 u cut2 Ascend 2 Ascend 1 Side1 Side2 A B Figure 4: Half display of a workpiece and two cutting tools. 5 PARIS: Reuse at Different Levels of Abstraction Paris (Plan Abstraction and Refinement in an Integrated System) [BW95] is a domain independent case based planning system which allows the flexible reuse of planning cases 5 PARIS: REUSE AT DIFFERENT LEVELS OF ABSTRACTION 10 cut u cut1 . Retrieved Case Current Situation Replay use left tool use right tool cut u cut2 use right ....

R. Bergmann and W. Wilke. Building and refining abstract planning cases by change of representation language. Journal of Artificial Intelligence Research, 3:53--118, 1995.

....of detail of the description of individual states can be reduced and second, the set of states that are kept in the abstract case can be reduced too. Formally, this transformation is decomposed into two independent mappings: a state abstraction mapping ff, and a sequence abstraction mapping fi [ Bergmann and Wilke, 1995 ] The state abstraction mapping transforms concrete state descriptions into abstract state descriptions, while the sequence abstraction mapping specifies which of the concrete states are mapped and which are skipped. Definition 1 (State Abstraction Mapping) A state abstraction mapping ff : S c ....

....is demonstrated in Figure 2. The concrete space shows the sequence of n operations together with the resulting state sequence. Selected states are mapped by ff into states of the abstract space. The mapping fi maps the indices of the abstract states back to the corresponding concrete states. In [ Bergmann and Wilke, 1995 ] we showed that in this abstraction methodology hierarchies of abstraction spaces as well as different kinds abstraction can be handled simultaneously. Moreover, our abstraction methodology is more powerful than abstraction by dropping conditions as often realized in hierarchical planning [ ....

[Article contains additional citation context not shown here]

R. Bergmann and W. Wilke. Building and refining abstract planning cases by change of representation language. Journal of Artificial Intelligence Research (to appear), 3:53--118, 1995. Also as Tech. Report LSA95-07E, Centre for Learning Systems and Applications, Univ. of Kaiserslautern.

....due to large spectrum of target problems processing of highly complex cases only concrete planning cases available (e.g. in archives of a company) 3. 2 Abstract Planning Cases We now summarize the formal definition of concrete and abstract planning cases already presented in detail in [ 5 ] . Representation of domains Following a STRIPS oriented representation [ 11 ] a planning domain D defines a (possibly infinite) set of states S and a set of operators O which describe transitions from one state to a successor state. A state s 2 S is described by a finite set of prepositions. A ....

....our intuition of case abstraction is captured in the following definition. A case C a is an abstraction of a case C c if there exists a state abstraction mapping ff and a sequence abstraction mapping fi, such that: s a j = ff(s c fi(j) holds for all j 2 f0; mg (see Fig. 5) In [ 5 ] we have discussed the generality of the presented case abstraction methodology. We have formally shown that hierarchies of abstraction spaces as well as abstractions with respect to different aspects can be represented using the presented methodology. Based on the defined levels of abstraction ....

[Article contains additional citation context not shown here]

R. Bergmann and W. Wilke. Building and refining abstract planning cases by change of representation language. Journal of Artificial Intelligence Research, 3:53--118, 1995.

....in which each edge leads from an abstract state to an abstract successor state. Finally, all consistent paths, starting at the abstract initial state and leading to the final abstract state are determined. Each of these paths represents a case which is an abstraction of the concrete case. In [ Bergmann and Wilke, 1995a ] we have described and analyzed this abstraction algorithm in detail and showed that it is sound and complete with respect to our abstraction methodology. 4 Retrieving and Refining Abstract Cases Now we assume a case base which contains a set of abstract cases that can be reused to solve a new ....

R. Bergmann and W. Wilke. Building and refining abstract planning cases by change of representation language. Technical Report LSA-95-07E, Centre for Learning Systems and Applications, University of Kaiserslautern, 1995.

No context found.

R. Bergmann and W. Wilke. Building and Refining Abstract Planning Cases by Change of Representation Language, Journal of Artificial Intelligence Research (JAIR), 1995, 3:53-118.

No context found.

Bergmann, R., & Wilke, W., Building and Refining Abstract Planning Cases by Change of Representation Language, Journal of Artificial Intelligence Research, Vol. 3, pp. 53-118, 1995.

No context found.

Bergmann, R. & Wilke, W. (1995a). Building and refining abstract planning cases by change of representation language. Journal of AI Research, 3:53--118.

No context found.

Building and refining abstract planning cases by change of representation language, Journal of AI Research, 3, pp. 53-118.

First 50 documents

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