K.F. Hammond, Case-based planning, Proceedings of Workshop on Case-Based Reasoning, (May 10 -- 13, 1988) pp. 17 -- 20.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....iterative. One example used to support the idea of case based reasoning is as follows: When an architect starts a new design for a client, he does not go back to first principles and try out all possible combinations. Instead he recalls past similar plans and modify them to fit his current needs [16]. R. Sun Fuzzy Sets and Systems 82 (1996 187 200 193 i have the following reservations about the purely symbolic case based approach. First of all, rules (i.e. abstract knowledge) are one of the most important cognitive mechanisms, although no doubt there are other kinds of knowledge (e.g. ....

K. Hammond, Case-Based Planning (Academic Press, San Diego, CA, 1989).

....selection, using a heuristic approach to find the minimal cost plan by merging separate plans. Once again, this is designed to find a plan that achieves multiple goals with minimal cost, as opposed to maximizing overall plan utility while allowing for partial goal failures. Case based planning (Hammond 1988; Alterman 1988) is primarily concerned with reusing plans from a library in order to increase the efficiency of plan generation. The key challenge is efficiently retrieving the most relevant plan and then adapting it to fit the current context. Our focus is on plan repair via merging, and the ....

Hammond, K. J. 1988. Case-based planning. In Kolodner, J., ed., Proceedings of Workshop on CaseBased Reasoning, 17--20.

....was based on Schank s dynamic memory theory and MOP theory. It was a question answer system with knowledge of the various travels and meetings of former US Secretary of State Cyrus Vance [4] Between 1983 and 1992, there appeared other systems which were based on Schank s theories, including CHEF [11], JULIA [13] CASEY [17] MEDIATOR [28] and PERSUADER [30] Between 1986 and 1989, Bruce Porter and his group, Concept Learning, at the University of Texas, Austin, studied the machine learning problem of concept learning for classification tasks. They developed the example based representation ....

Kristian J. Hammond. Case-based planning. Academic Press, 1989.

....only the problem description) In the following subsections I will review the models proposed for retrieval and mapping in more detail. 7.2.1. Retrieval. There are several main approaches to retrieval. 1) In case based reasoning (Carbonell, 1983, 1986, Kolodner and Simpson, 1986, 1989, Hammond, 1989, 1990) retrieval is performed on the basis of a specific organization of LTM around an indexing scheme. Thus Carbonell (1986) indexes the potential sources of analogy by the first reasoning steps in the problem solving activity of that case (he called this case derivational analogy) Kolodner ....

.... of that case (he called this case derivational analogy) Kolodner Simpson (1986) use a specific organization of previous cases organized around a generalized episode (MOP, EMOP) and indexed by their differentiating features (this organization could be considered as a discrimination net) Hammond (1989, 1990) indexes old plans by goals and the problems that they avoid and in this way the basic organization of the plan memory is a discrimination net. In this way each particular model has its specific memory organization (specific indexing scheme) depending on the task it solves. This can be ....

Hammond, K. (1989). Case-Based Planning. Boston, MA: Academic Press.

....make robot programming even more accessible to novice users, without compromising the expressiveness of the graphical programming language. One opportunity for expanding our model of planning further is to introduce the idea of storing and reusing plans as in case based planning [Alterman, 1988; Hammond, 1988] As with human programmers, who patch together a plan using routines and plan fragments that achieve subpieces of a plan, the automatic planning techniques could store plan pieces that achieve specific goals and reassemble them to form new plans [Laird et al. 1987; Redmond, 1990] Also, given ....

K. J. Hammond. Case-based planning. In J. Kolodner, editor, Proceedings of Workshop on Case-Based Reasoning, pages 17--20, May 1988.

....(b) fly from Auckland to Calgary, and (c) go from Calgary to Banff. A planning system that reasons about the primitive actions (e.g. go forward, turn right, turn left) will quickly be overcome by the tyranny of detail. Case based planning reuses previous solutions when solving new problems [Ham89]. Given a problem, a similar plan is retrieved and adapted to the new situation. The adaption methods are specific plan debugging routines. The assumption is that a similar plan can be found efficiently and that the cost of adapting an old solution is less than that of creating one from scratch. ....

....point of view, operator insertion and removal are the only ones necessary, since any plan can be created with these two transformations alone. However, this neglects the fact that these transformations have specific conditions under which they are applied. For example, in Hammond s Chef planner [Ham89], an operator can only be replaced if it solves a missing precondition or unwanted side effect conflict. As Hammond showed, the power of Chef stems from the fact there is a small set of these transformations and applicability conditions that can solve most problems in the cooking domain. 4 ....

Kristian J. Hammond. Case Based Planning. Academic Press Inc., 1989.

.... nearest neighbour to the current case, and the neighbour s solution is used directly (e.g. KA88] In more complex manifestations of CBR, there may be a more complex matching process and various means of modifying the solution of the previous case before applying it to the new situation (e.g. Ham89a, HK91] In addition to CBR other methods of EBR can be identified. 7 ffl Analogical Reasoning: Analogical reasoning attempts to transfer knowledge from one domain (the source domain) to another (the target domain) Pri88] The knowledge required to perform this operation is an analogical ....

....case. In order to confirm a choice some element of domain knowledge may be attached to the previous cases to verify the selection. This method is used by the CASCADE [SM91] system. 8 ffl Adaptation of cases may require some knowledge of the domain to be encoded as in the CHEF adaptive planner [Ham89a] Our distinction is further blurred by considering two research strands which seek to exploit the strengths of both reasoning styles. For example in Derivational Analogy [Car86] the system may have sufficient domain knowledge to (in principle) solve any problem from the domain model alone. Case ....

Kristian Hammond. Case Based Planning. Acedemic Press, 1989.

....based rules for individualized tutoring processes. Summing up, there are two major problems we have to deal with. 1. Acquisition of planning rules for individualized tutoring. 2. Pursuing a set of interacting subgoals. Case based planning is especially suitable for both of these problems [5]. Instead of creating new plans from scratch, it supports the reuse of old plans, which pursue a set of subgoals and may be adapted by modification rules. The repaired plans can be stored and used for an optimized planning in new situations. We express tutoring plans by thread configurations. For ....

Hammond K.: Case-Based Planning. Academic Press, INC., Boston (1989).

....for adapting previous solutions. Explanation occurs when a previous solution has some flaw or when a flaw is detected while adapting a previous solution. By explaining the flaw, the system can learn to avoid that type of failure in the future: this is the essence of part of the way Hammond s CHEF [17, 16] case based planner operates. 2.2 Integration of communications In the original CASSANDRA design [2] we described a communications architecture in terms of channels and ports, but did not investigate a protocol or language for communications. We assumed that we would be able to do this at a ....

....experience in performing its various tasks. We view each agent as being able to perform a number of tasks so that it can flexibly respond to the environment. For reasons that we gave in the last section, the problem solving component in the new model is a Case Based Reasoning (CBR) system [25, 26, 17, 24, 20]. CBR, we believe, provides both flexibility and the ability to re use previous experience in solving problems. In addition, CBR is a way of constructing and understanding systems that learn [25, 26] The central component of a Case Based Reasoning system is a dynamic memory: this memory is ....

Hammond, K. J., Case-Based Planning, Academic Press, Boston, 1989.

.... Controller Card Transformer Wiring A 3 A 4 P 6 P 7 A P 8 P 9 P 2 Disk Drives Controller Card PCBs Subassembly Floppy Disk Subassembly Hard Disk Subassembly Hard Disk Housing A 0 P 1 A 1 P 5 A 2 P 4 P 11 P 10 A 6 P 3 A 3 A 4 P P 7 A 5 P P P 2 s[1] s[2] s[3] s[5] s[6] s[4] a) b) Figure 3. Disassembly tree representation of a PC computer. The most prominent application of modularity has come from the computer industry. With an enormous growth of complexity in the design of computers (and other electronic products) the developers manage these by ....

....modules and decomposed into simple sub DTs that define the subassemblies. As an example, consider the DT representation a PC computer shown in Figure 3. There are eleven parts in the PC. The parts can be clustered into six subtree modules, four of which are simple subtrees (viz. s[2] s[4] s[5], and s[6] and two are complex subtrees (viz. s[1] and s[3] In this representation, the structure of complex subtree s[1] is composed of subtrees s[2] s[3] and s[6] The complex subtree s[3] is composed of two simple subtrees, s[4] and s[5] Veerakamolmal and Gupta, 1998a] Disassembly ....

[Article contains additional citation context not shown here]

Hammond, K. J., Case-Based Planning, Academic Press Inc., San Diego, CA, USA, 1989.

....experiences for both of these processes, we are employing case based reasoning . Case based reasoners solve problems or examine and explain possible outcomes to a scenario by relying on prior similar experiences. Case base reasoning (CBR) systems can be found in such diverse domains as cooking [8], medical diagnostics [11] manufacturing [10] game playing [12, 4] and legal applications [1, 15] This research was supported while the first author was at the University of Massachusetts at Amherst by NSF Grant no. EEC9209623, State Industry University Cooperative Research on Intelligent ....

Kristian J. Hammond. Case-Based Planning. Academic Press, Inc., 1989.

....can be completely anticipated, we possess a vast collection of more general knowledge that allows us to integrate alternatives seamlessly with our routine. We divert from our routine when it makes sense to do so, and return to it without anything like the kind of effort known to be required (e.g. [11]) to alter a stored symbolic plan. We can also use our routine as a weaker guide in conjunction with background knowledge to cope in a satisficing manner with even greater degrees of variation. For example, one can shop reasonably successfully even when in a hurry and in a strange supermarket; can ....

Hammond, Kristian, Case-Based Planning (Boston: Academic Pr.), 1989. 277 pp.

....plan revision. 7 Related Work Plan revision techniques can be classified into ones that reason about the structure of plans and the dependencies between plan steps [Sus90, Kam90, HW92] and others that revise plans based on simulations projections of plans (Hacker [Sus90] Gordius [Sim92] Chef [Ham89] There are several differences between fppd and other plan revision techniques: 1) fppd differs from other transformational planners in that it tries to adapt a simultaneously executed plan to specific situations to avoid particular flaws and thereby improve the agent s behavior instead of ....

K. Hammond. Case-Based Planning. Academic Press, Inc., 1989.

....can [MR91] We have implemented models for about twenty five types of behavior flaws and about forty transformation rules. Most of which cannot be expressed in other planning representations. xfrml can easily express the rules used by other transformational planners, such as those described in [Sus77, Ham89]. To implement gordius like plan transformations, xfrm s projector would have to construct a dependency structure that justifies the executability of plan steps and their effects. We don t know how much of rpl could be handled by such dependency structures because the identification of causes and ....

K. Hammond. Case-Based Planning. Academic Press, Inc., 1989.

....a vast collection of more general knowledge that allows us to integrate alternatives seamlessly with our routine. We can divert from our routine when it makes sense to do so, and return to it without anything like the kind of effort it is known to require to alter a stored symbolic plan (e.g. [Hammond, 1989]) Beyond reacting to the unexpected, we can use this same background knowledge in conjunction with our usual routine to cope in a satisficing manner with even greater degrees of variation. For example, one can prepare a meal for a larger number of people, with different equipment than one is ....

Hammond, Kristian, Case-Based Planning (Boston, MA: Academic Press), 1989. 277 pp.

....this endeavor. For both of these processes we are employing case based reasoning . Case based reasoners solve problems or examine and explain possible outcomes to a scenario by relying on prior similar experiences. Case base reasoning (CBR) systems can be found in such diverse domains as cooking [21], medical diagnostics [32] manufacturing [26] game playing [36, 37, 9] and legal applications [54, 58, 57] In the legal domain, finding similar cases is crucial because of the importance precedents play when arguing a case. By examining the outcomes and features of previous cases, a legal ....

....not search the entire case base for the best matches [32] Many CBR systems use some form of nearest neighbor metric. Other methods include claim lattice, as implemented in HYPO [54, 2] structure mapping as implemented in GREBE [4] or Thematic Organization Packets (TOPs) as implemented in Chef [21]. In SPIRE, we use a HYPO style CBR module that employs a claim lattice to determine similarity. A claim lattice is a partial ordering of the cases similar to the cfs. The CKB is sorted based on the intersection of each case s dimensions (or features) with those applicable to the cfs; cases with ....

[Article contains additional citation context not shown here]

Hammond, Kristian J. Case-Based Planning. Academic Press, Inc., 1989.

....it can build a domain knowledge of its own. As discussed in Section IX C, such knowledge can be used to assist the action recognition and the task execution. One possibility would be detecting failures and or generating local patch plans based on similar past experiences using Case Based Reasoning[31]. In order to realize this, an appropriate action memory structure must be explored. C. Treatment of Errors and Failures C.1 Positional Errors. Proper treatment of positional errors is crucial for successful assembly tasks. The presented framework contains three major source of errors: visual ....

K. J. Hammond, Case-Based Planning, Academic Press, 1989.

....problems. She stresses that this model for design synthesis requires design episodes in the form of cases rather than generalisations about a design domain as in decomposition (Maher, 1990) Case based reasoning involves several operations and these have been well documented in the literature (Hammond, 1989; Kolodner, 1991; Reisbeck and Schank, 1989) Briefly, they include, retrieving the relevant cases from case memory (retrieval) selecting the most promising case (mapping) modifying its solution for use in the new problem situation (adaptation) testing the solution, evaluating the results, ....

....conflicts into the design that must be resolved; a specialist that alters one feature of a design may effect another feature. To this end, we propose a second type of adaptation model, strategies. Each strategy corresponds to some generalised conflict configuration that can occur; Hammond (Hammond, 1989) outlines a number of such failure configurations that are applicable to a range of planning and design tasks. Each strategy also contains a number of repair methods that can resolve the conflict in question. Returning to the GUI design domain one common conflict configuration occurs when the ....

Hammond, K.J. (1989). Case-Based Planning, Academic Press.

....when it has allocated resources to an activity. The check is done by going through the already scheduled activities and their allocated resources, and comparing the total allocated resources with the maximum resources available. The planning described here differs from ordinary AI planning ([AC87, Geo88, GL87, GI89, Kae86, Kae88, Wil89, BD90, Ham89, GN87] are some papers giving a short survey) where the system tries to find a sequence of actions that leads to a desired state based on descriptions of different actions, which are performed in more (classical planning) or less systematic ways (reactive planning) The more systematic approaches often ....

....systematic approaches often have something that guides the planning algorithm to its desired state in the form of a leads to operator [Kae88] or something similar, but seldom do these system have methods or canned plans based on experience. One exception are so called case based planning systems [Ham89], but in such systems the planning system itself tries to deduce its experiences from failed plans, and not in cooperation with a user. In general, AI planning systems try to solve the problem of planning for an autonomous agent, which is something else than giving a user support for making plans, ....

K. J. Hammond, Case-Based Planning, Academic Press, 1989

No context found.

K.F. Hammond, Case-based planning, Proceedings of Workshop on Case-Based Reasoning, (May 10 -- 13, 1988) pp. 17 -- 20.

No context found.

Kristian J. Hammond. Case Based Planning. Academic Press Inc., 1989.

No context found.

Hammond, K. J. 1989a. Case-based planning. Boston: Academic Press.

No context found.

Hammond, K. (1989) Case-Based Planning. Academic Press.

No context found.

Hammond, K. (1989a). Case Based Planning. Academic Press.

No context found.

Germany. Hammond, K. J. (1989). Case-based Planning. San Diego, CA: Academic Press.

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