Schank, R.C.: Explanation Patterns: Understanding Mechanically and Creatively. Lawrence Erlbaum Associates, Hillsdale, NJ (1986)  Home/Search   Document Not in Database   Summary   Related Articles   Check

....to improve its ability to reason and learn. 4 A taxonomy of reasoning failures Based on the above architecture, we can characterize the types of reasoning failures that a reasoner might encounter. The term reasoning failure includes not simply performance errors, but also expectation failures [Schank, 1986], anomalous situations which the reasoner failed to predict, and other types of reasoning failures as well. By using failures to guide learning, the reasoner can focus on the problems that actually arise rather than searching through the more complex space of problems that can theoretically arise ....

R.C. Schank. Explanation Patterns: Understanding Mechanically and Creatively. Lawrence Erlbaum Associates, Hillsdale, NJ, 1986.

....Or a committee may make a recommendation which may or may not be accepted by the higher authority. In these cases, the explanation plays a central role in effecting the decision. We note that this type of explanation is different from the usual sense of explanation found in the CBR literature [Schank, 1986, Ram, 1989, Kass, 1990, Leake, 1990] Previous researchers have focused on explanation of anomalous observed events as part of the process of learning. Our present use of explanation is complementary to that process: decision makers offer explanations for the benefit of observers who may find the ....

Schank, R. (1986). Explanation Patterns: Understanding Mechanically and Creatively. Lawrence Erlbaum Associates, Hillsdale, NJ.

....advisory system must also be accountable. It must be able not only to offer advice, but also to justify its position by providing an explanation of its reasoning. We note that this type of explanation is different from the usual sense of explanation found in the case based reasoning literature [7, 4, 5]. Previous researchers have focused on explanation of anomalous observed events as part of the process of learning. Our present use of explanation is complementary to that process: decision makers offer explanations for the benefit of observers who may find the decision to be anomalous. We can ....

R.C. Schank. Explanation Patterns: Understanding Mechanically and Creatively. Lawrence Erlbaum Associates, Hillsdale, NJ, 1986.

....The political science literature (Kingdon 1973, Fenno 1978) indicates that members of Congress are cognizant of the role of explanation in decision making. We note that this type of explanation is different from the usual sense of explanation found in the casebased reasoning literature (Schank 1986, Ram 1989, Kass 1990, Leake 1990) Previous researchers have focused on explanation of anomalous observed events as part of the process of learning. Our present use of explanation is complementary to that process: decision makers offer explanations for the benefit of observers who may find the ....

R.C. Schank. Explanation Patterns: Understanding Mechanically and Creatively. Lawrence Erlbaum Associates, Hillsdale, NJ, 1986.

....a recommendation which may or may not be accepted by the higher authority. In these cases, the explanation plays a central role in effecting the decision. We note that this type of explanation is different from the usual sense of explanation found in both the case based reasoning literature [ Schank, 1986, Ram, 1989, Kass, 1990, Leake, 1990 ] and the explanation based learning literature [ DeJong, 1981, DeJong, 1985, Mitchell et al. 1986, Ellman, 1989 ] Previous researchers have focused on explanation of anomalous observed events as part of the process of learning. Our present use of ....

R.C. Schank. Explanation Patterns: Understanding Mechanically and Creatively. Lawrence Erlbaum Associates, Hillsdale, NJ, 1986.

....the best we can do is provide the most likely explanation, not a correct one. Our inability to deductively reason about the intention of a cycle is not so grim as it might seem. People routinely make such inferences in the social domain without a complete model, a fact that Schank has stressed [45]. Providing a plausible answer rapidly is often quite valuable. We contend that the evidential reasoner s lack of deductive soundness, when made explicit to students, is useful pedagogically, in that it requires the student to pass judgment on the validity of the advice. The system s explanations ....

R.C. Schank, Explanation Patterns: Understanding Mechanically and Creatively, Lawrence Erlbaum Associates, Hillsdale, NJ, 1986.

....out precisely those places where the system most needs learning: places where its current knowledge or reasoning is inadequate. Failures are therefore viewed as opportunities for learning which will improvetheperformance of the system #Leake, 1992; Krulwich et al. 1992; Hammond, 1989; Ram, 1989; Schank, 1986; Riesbeck, 1981#. Focusing on failures simpli#es the introspective reasoner s task, and concentrates e#ort on those places where opportunities to learn are most strongly indicated. If the underlying reasoning appears perfect, there is no reason to expend e#ort secondguessing it. Model based ....

Schank, R. #1986#. Explanation Patterns: Understanding Mechanically and Creatively. Lawrence Erlbaum Associates, Hillsdale, NJ.

....of imperfect knowledge, limited reasoning resources, and pragmatic motivations for explaining. An alternative model explanation generation relying on case based reasoning was developed largely to address the problems of everyday abductive explanation #Kass, 1986; Leake, 1992; Leake Owens, 1986; Schank, 1986; Schank, Riesbeck, Kass, 1994#. The case based model generates new explanations by retrieving explanations of relevant prior episodes and adapting them to #t the new situation in light of the explainer s needs for information. In this model, the prior experiences of the explainer are ....

....given the likelihood that #aws will exist in the explainer s domain theory. Consequently, the case based model explicitly treats explanations as plausible inference chains rather than deductive proofs. In the case based explanation model, explanations are represented as explanation patterns #XPs# #Schank, 1986#. Explanation patterns encode chains of belief dependencies showing how belief in a conclusion follows from belief in a set of premises. Syntactically these are like deductive explanations, but the derivation in an explanation pattern is not considered to entail the chain s consequent. Instead, ....

[Article contains additional citation context not shown here]

Schank, R. #1986#. Explanation Patterns: Understanding Mechanically and Creatively. Lawrence Erlbaum Associates, Hillsdale, NJ.

....system helps the student to explore the ramifications of changing some aspect of an animal. Each CreANIMate dialogue is based around a question that is fundamental to understanding the ways in which animals survive in the wild. We call these questions explanation questions, in the terminology of Schank (1986). Explanation questions are the questions in any domain that a knowledgeable individual asks to construct an explanation for a phenomenon in that domain. Some explanation questions in animal morphology are, Why is it useful for this animal to perform a particular action and How does this ....

Schank, R.C. 1986. Explanation Patterns: Understanding Mechanically and Creatively . Hillsdale, NJ: Lawrence Erlbaum Associates.

.... [5] abduction underlies numerous processes of a great practical importance, such as medical diagnosing [2, 20, 57, 62, 68, 70, 83] testing and repairing technical devices and structures [4, 11, 13, 21, 24, 40, 55, 58, 72] planning a course of action [7, 18, 59] natural language understanding [32, 33, 43, 44, 39, 74, 75], learning [14, 61, 74, 75] Let S denote a Knowledge System (presented in the language of First Order Logic) that describes a part of a real world W . Suppose that a state Obs has been observed in W , and we wish to figure out what event or state of affairs caused Obs or accounts for its ....

.... of a great practical importance, such as medical diagnosing [2, 20, 57, 62, 68, 70, 83] testing and repairing technical devices and structures [4, 11, 13, 21, 24, 40, 55, 58, 72] planning a course of action [7, 18, 59] natural language understanding [32, 33, 43, 44, 39, 74, 75] learning [14, 61, 74, 75]. Let S denote a Knowledge System (presented in the language of First Order Logic) that describes a part of a real world W . Suppose that a state Obs has been observed in W , and we wish to figure out what event or state of affairs caused Obs or accounts for its occurrence. To do so we have to ....

Schank, R. C., 1986, Explanation Patterns: Understanding Mechanically and Creatively. Lawrence Erlbaum, Hillsdale, N. J.

....while rules can at best be conjectured. One of the theoretical advantages of this approach is that, while rules tend to give rise to inconsistencies, cases cannot contradict each other. Our approach is closely related to (and partly inspired by) the theory of CaseBased Reasoning (CBR) proposed by Schank (1986) and Riesbeck and Schank (1989) See also Kolodner and Riesbeck (1986) and Kolodner (1988) In this literature, CBR is proposed as a better AI technology, and a more realistic descriptive theory of human reasoning than rule based models (or systems) However, our approach differs from theirs in ....

Schank, R. C. (1986), Explanation Patterns: Understanding Mechanically and Creatively. Hillsdale, NJ, Lawrence Erlbaum Associates.

....today s prediction problem and past observations. That prediction is based on past cases is beyond doubt. Over the past decades Hume s approach has found re incarnations in the artificial intelligence literature as reasoning by analogies, reasoning by similarities, or case based reasoning. See Schank (1986) and Riesbeck and Schank (1989) Many authors therefore accept the view that analogies, or similarities to past cases hold the key to human reasoning. Moreover, the literature on machine learning and pattern recognition deals with using past cases, or observations, for predicting or classifying ....

Schank, R. C. (1986), Explanation Patterns: Understanding Mechanically and Creatively. Hillsdale, NJ: Lawrence Erlbaum Associates.

....d d safe , Postconditions Success 9 : l 2 l light (if car 2 has driven past the light) slow down before red light, alone will be activated. Failure: car 1 collides into car 2 3. 6 Knowledge Representation Issues XPlans are inspired in part by Schank s Explanation Patterns [Schank, 86] However, they present some substantial differences, since the context we are working in is entirely different from Schank s. We have to consider an evolving value for car 2 D 2 , are used. In the recognition algorithm, the measured value D 1 is compared with the ideal value d. 9 Postconditions ....

....input and lacunary input, although these questions are classic in statistical signal recognition and in the machine learning community (see for instance [Kodratoff et al. 90] So each of the following works only covers a part of the requirements defined above. Schank s Explanation Patterns [Schank, 86] describe the possible motivations, hence justifications, about normal situations. In the case of abnormal events, a tweaking mechanism performs adjustment of the existing relevant plan so that it will cover the new case. General knowledge on motivations is expressed in the form of common sense ....

Schank R.C., Explanation Patterns: Understanding Mechanically and Creatively, Lawrence Erlbaum Associates, 1986.

....merging three kinds of different knowledge, general principles of tactics, intelligence information, and the doctrine of both sides. In order to merge these three kinds of information in a retrievable form, we had to develop a special knowledge representation framework, inspired from Schank s XPs (Schank, 1986; Schank and Kass, 1990) in which a special part is devoted to the slow emergence of a plan when some partial information confirms its activation. 7 Use ML to acquire knowledge usually compiled by experts 7.1 Acquiring perceptual chunks Example 9. Solving geometry problems. This is a ....

Schank R., Explanation Patterns: Understanding mechanically and creatively, Lawrence Erlbaum 1986.

....content and representation of these meta models is presented along with a computer model, Meta AQUA, that implements this theory for a story understanding task. The theory focusses on failure driven learning. The term failure includes not simply performance errors, but also expectation failures (Schank, 1986), anomalous situations which the reasoner failed to predict, and other types of reasoning failures as well. Unlike successful processing where there may or may not be anything to learn, failure situations are guaranteed to provide a potential for learning, otherwise the failure would not have ....

....which provide conceptual coherence to the story by linking the pieces of the story together. The approach to explanation construction is case based, and is based on Schank s theory of explanation patterns (XPs) in which explanations are built by applying known XPs to the events in the story (Schank, 1986; Ram, 1990a) Expectation failures arise when the world differs from the system s expectations. For example, the system may be faced with an anomalous situation in which the XP that the system believes to be applicable turns out to be contradicted in the story. When the system encounters an ....

[Article contains additional citation context not shown here]

Schank, R.C.. Explanation Patterns: Understanding Mechanically and Creatively, Lawrence Erlbaum Associates, Hillsdale, NJ, 1986.

.... is not large, and itself says nothing about recognition from events, or from noisy input and lacunary input, although these questions are classic in statistical signal recognition and in the machine learning community (see for instance [Kodratoff et al., 90] Schank s Explanation Patterns [Schank, 86] aim at describing the possible motivations of agents and at accounting for the mechanisms of creativity as robust explanation devices which track causal chains. If the situation cannot be explained in a straightforward way, a tweaking mechanism performs an adjustment of the existing relevant ....

Schank R.C., Explanation Patterns: Understanding Mechanically and Creatively, Lawrence Erlbaum Associates, 1986.

....it did not generate that expectation prior to encountering the sentence containing a person that struck the ball. To satisfy the comprehension task in complicated or unusual input, questions may be raised and an explanatory process may be invoked. Hence, the expected outcome will be an explanation [72,74,88,99]. 3 That is, the reasoner consciously anticipates a certain explanation to be true of some object or event in the input that is to be explained. When these explanations prove incorrect, such as the explanation that Lynn wanted to hurt the ball, explicit expectations can be violated as well. ....

....about its own memory system when it forgets and can reason about its knowledge and inferences when it draws faulty conclusions. 2.3.1. Representing contradiction: A symptom of failure We posit a declarative representation for mental state transformations using explanation pattern (XP) theory [73,74,88,89,91] and call the representational structures meta explanation patterns (Meta XPs) Because meta X literally means X about X, a meta explanation pattern is an explanation pattern about another explanation pattern. While a standard XP is a causal structure that explains a physical state by ....

R. C. Schank, Explanation Patterns: Understanding mechanically and creatively (Lawrence Erlbaum Associates, Hillsdale, NJ, 1986).

....vocabulary of planning concepts. Proverbs are culturally shared cases : they identify generic strategies that everyone uses to deal with commonly occurring problems in planning and social interaction. As such, proverbs provide a rich source of data on culturally shared models of planning [ Schank, 1986; White, 1987 ] Different proverbs implicitly presuppose different culturally shared models. Representing a large number of proverbs has proved an effective strategy for developing and testing our representational vocabulary. It is, however, important to emphasize that we have no special ....

....because any action that can be represented in the external vocabulary can also be redescribed in terms of a planning action in the canonical model. The canonical model describes planning in terms of plan design and plan execution, building on Schank s idea of the goal plan action state chain [ Schank, 1986 ] Third, there is the operational vocabulary of the planner s data structures, in which the outputs of the adapter are encoded. EXTERNAL ADVICE REPRESENTATIONS OPERATIONAL ADVICE (Planner data structures) CANONICAL KERNEL 3.Translation 2.Causal reasoning 1.Translation Figure 1: Information ....

Schank, Roger C. 1986. Explanation Patterns: Understanding Mechanically and Creatively. Lawrence Erlbaum Associates, Hillsdale, NJ.

....and Ortony, 1992; Elliott and Siegle, 1993; Elliott, 1993; Elliott, 1994b ] Motivation Overview All stories need a point . This will be just as true of interactive fiction as it is of static fiction. One common view is that the point of a story arises largely from failed expectations (c.f. Schank, 1986; Schank, 1990 ] It is possible, however, that a computational model of stories might be mislead by failed expectations which do not yield stories (e.g. the weather was hot for four straight days in October) and by stories which actually seem enhanced when our expectations are met (e.g. ....

R.C. Schank. Explanation Patterns: Understanding Mechanically and Creatively. Lawrence Erlbaum Associates, Hillsdale, NJ, 1986.

....basic problem understanding model. A conflict case integrates several pieces of explanatory knowledge whereby each describes a particular point of view that accounts for the emergence of a conflict. The explanatory knowledge used in this work is based on the notion of Explanation Pattern (XP) 1 (Schank, 1986). After the system explains the new conflict case by using several XPs, it uses the XPs to retrieve several sets of cases where each set contains a number of cases indexed under each XP. Old cases which share the largest number of XPs with the new case are preferred over cases with less XPs (given ....

Schank, R. C. (1986); Explanation Patterns: Understanding Mechanically and Creatively; Lawrence Erlbaum Associates, Hillsdale, New Jersey.

....such a failure in the future. Explaining anomalies is prevalent in all of our problem solving and understanding activities. Explanation has been called the credit assignment problem and the blame assignment problem, depending whether it explains a success or a failure. A case based explanation (Schank, 1986; Ram, 1993b; Ram, 1993a) explains a phenomenon by remembering a similar phenomenon, borrowing its explanation, and adapting it to fit. In this context, CBR fits very naturally and will improve the explanation process. A case based explanation requires the following mechanisms: ffl a retrieval ....

Schank, R. C. (1986). Explanation Patterns: Understanding Mechanically and Creatively. Lawrence Erlbaum Associates, Hillsdale, NJ.

....generating hypotheses in this way are, first, to characterise the problem in a way that makes the old explanations come to mind, and second, to tweak the old explanations into something that fits the new situation. A full presentation of our current work is clearly impossible in this paper (see [Schank 86] for a theoretical discussion and [Kass 86] and [Leake and Owens 86] for brief discussions of a program built around these .principles) the goal here is simply to point out how our interest in natural language processing has led us naturally, and indeed inevitably, to develop theories of ....

Schank, R.C., Explanation Patterns: Understanding Mechanically and Creatively, 1986. Book in press. 115

No context found.

Schank, R.C.: Explanation Patterns: Understanding Mechanically and Creatively. Lawrence Erlbaum Associates, Hillsdale, NJ (1986)

No context found.

R.C. Schank. Explanation Patterns: Understanding Mechanically and Creatively. Lawrence Erlbaum Associates, Hillsdale, NJ, 1986.

No context found.

Schank, R. (1986). Explanation Patterns: Understanding Mechanically and Creatively. Lawrence Erlbaum Associates, Hillsdale, New Jersey.

First 50 documents  Next 50

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