H. A. Kautz. A Formal Theory of Plan Recognition. PhD thesis, University of Rochester, 1987.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....undo mechanism to minimally change hypotheses and does not generate and consider all possibilities. Lastly, BELIEVER s plan recognition did not support interleaved plans as ours does. 3. 2 Kautz Generalized Plan Recognition The other foundation work in plan recognition was that of Kautz [KA86, Kau87, Kau90, Kau91] Kautz cast plan recognition as the logical inference process of circumscription. This logical cast on plan recognition allowed him to use the a very rich knowledge representation (essentially that of first order logic) as well as to use temporal logic to represent actions and ....

Henry A. Kautz. A formal theory of plan recognition. Technical Report 215, University of Rochester, Department of Computer Science, 1987. PhD Thesis.

....actions. Four principal methods for plan recognition have been proposed in the literature. The methods are plausible inference (Allen [1] Carberry [2] Litman [15] Sidner [25] parsing (Huff and Lesser [9] circumscribing a hierarchical representation of plans and using deduc tion (Kautz [12, 13]) and abduction (Charniak and McDermott [6] Konolige and Pollack [14] Poole Our particular interest is in the use of plan recognition in question answering systems, where recognizing the plan underlying a user s queries aids in the generation of an appropriate response. Here, the plans and ....

....which is being pursued. Below is an algorithm that determines what to say. Our algorithm for estimating whether the ambiguity matters is not dependent on the method of plan recognition used. Now our proposal for clarification dialogues is tied to a hierarchical plan library in the style of Kautz [12]. The input to the algorithm is a set of possible plans where the user s action is related to top level or end goals through chains of goals. Each plan in the set is annotated with a a critique. The key idea is to ask about the highest level possible, check whether the ambiguity still needs to be ....

[Article contains additional citation context not shown here]

Kautz, H. A. 1987. A Formal Theory of Plan Recognition. Ph.D. thesis available as University of Rochester Technical Report 215, Rochester, N.Y.

....undo mechanism to minimally change hypotheses and does not generate and consider all possibilities. Lastly, BELIEVER s plan recognition did not support interleaved plans as ours does. 3. 2 Kautz Generalized Plan Recognition The other foundation work in plan recognition was that of Kautz [KA86, Kau87, Kau90, Kau91] Kautz cast plan recognition as the logical inference process of circumscription. This logical cast on plan recognition allowed him to use the a very rich knowledge representation (essentially that of rst order logic) as well as to use temporal logic to represent actions and ....

Henry A. Kautz. A formal theory of plan recognition. Technical Report 215, University of Rochester, Department of Computer Science, 1987. PhD Thesis.

....use plan recognition. This observation was among the beginnings of the planrecognition field. The implementation, however, was never fully implemented nor specified. Kautz Generalized Plan Recognition The other foundational work in plan recognition was that of Kautz ( Kautz and Allen, 1986; Kautz, 1987; Kautz, 1990; Kautz, 1991] Kautz cast plan recognition as the logical nonmonotonic inference process of circumscription. This logical cast on plan recognition allowed him to use the a very rich knowledge representation (essentially that of first order logic) Kautz represented the space of ....

Henry A. Kautz, "A Formal Theory of Plan Recognition," Technical Report 215, University of Rochester, Department of Computer Science, 1987, PhD thesis.

....unsupervised learning algorithm for adapting a recognizer to a person s idiosyncratic behaviors. Our experiments in two domains show that applying Adapt to the BOCE recognizer can improve its performance by a factor of two to three. 1 Introduction Goal recognition (e.g. Kautz, 1987; Carberry, 1990 ] is the task of inferring a person s intentions given a partial view of their actions. As noted in [ Maes and Kozierok, 1993; Bauer, 1994; Ardissono and Cohen, 1995 ] goal recognition is difficult, in part, because people do not all act alike. The same actions can warrant ....

....remove a precondition or effect from the domain operators. This could potentially improve recognition by allowing the goal recognizer to build plans based on what preconditions and effects the observed person believes actions have. In the widely used framework for plan recognition described in [ Kautz, 1987 ] one possible adaptation is to add or remove an abstraction link between nodes in the plan hierarchy. Another adaptation in Kautz s framework is adding or removing ordering constraints between steps in the plan hierarchy. Finally, the notion of background goals could be applied to almost any ....

[Article contains additional citation context not shown here]

H. Kautz. A Formal Theory Of Plan Recognition. PhD thesis, University of Rochester, 1987.

....specific knowledge in order to gain a computational speedup. There are mainly two different advances to domain independent plan recognition. 3.1 Deduction Approach, Kautz 1987 The first attempt to examine plan recognition in a more general context was made by Henry Kautz in his Ph.D. thesis [13]. The theory he presented was the 1 GUIDON II tried to teach students how to make meningitis diagnosis and used the same knowledge base as MYCIN [3] one of the first expert systems) MYCIN, developed at Stanford in the mid 1970, was one of the first programs to deal with the problems of reasoning ....

.... Marinara Any Event Prepare Meal Wash Dishes Make Pasta Dish Make Spaghetti Pesto Make Spaghetti Marinara Make Fettucini Alfredo Boil Make Noodles Make Sauce Make Meat Dish Make Fettucini Make Pesto Make Alfredo Sauce Make Spaghetti Make Marinara Figure 1: A simple cooking hierarchy (after [13]) End Prepare Meal Make Pasta Dish Make Meat Make Spaghetti Marinara Make Chicken Marinara Make Marinara End Make Noodles Make Pasta Dish Prepare Meal End Prepare Meal Make Pasta Dish Make Spaghetti Marinara Make Noodles Make Marinara results in the following explanation graph: ....

[Article contains additional citation context not shown here]

Kautz, H. (1987). A Formal Theory of Plan Recognition. TR 215 Department of Computer Science, University of Rochester.

....it is explained how heuristics can reduce the search space for possible explanations resulting some observed input actions. Results gained from the implementation are used to motivate future activities. 1 A Reliable Computer Human Interaction Environment 2 A Formal Theory of Plan Recognition [15] Declaration I declare that the special study described in this dissertation has been carried out and the dissertation composed by me, and that the dissertation has not been accepted in fulfillment of the requirements of any other degree or professional qualification. Stefan Huwer Certificate ....

....of plans that are possible in the application domain. Therefore a change of the application domain results only in modifying the plan library. 2.3.1 Deduction Approach, Kautz 1987 The first attempt to examine plan recognition in a more general context was made by Henry Kautz in his Ph.D. thesis [15]. The theory he presented was the first really theoretical approach to the problem of generalised plan recognition. 2 GUIDON II tried to teach students how to make meningitis diagnosis and used the same knowledge base as MYCIN [3] one of the first expert systems) MYCIN, developed at Stanford in ....

[Article contains additional citation context not shown here]

Kautz, H. (1987). A Formal Theory of Plan Recognition. TR 215 Department of Computer Science, University of Rochester.

....les symptomes sont multivalu es : il suffit d imaginer ce qui arrive quand on enonce qu en raison d une panne un grand nombre de steps peuvent se produire dans les prochaines 10 minutes. Param etrer les hypoth eses devient r eellement difficile. Dans le contexte de la reconnaissance de plan, [Kautz 1987] s appuie sur une hi erarchie d ev enements (liens d abstraction et de d ecomposition) et consid ere tous les emplois d un fait observ e avant de l expliquer par abduction (la relation emploi etant grosso modo l inverse de la relation est un composant de ) Preist 1994] propose d appliquer ....

H. A. Kautz. A formal theory for plan recognition. Univ. of Rochester, TR-215, 1987.

....in an ideal context, where the input is supposed to be 100 correct and consists of actions, that is, chunks of activity representing well defined transitions in the situation. This gives rise to strong theory domains to which it is natural to apply a formal treatment, such as abduction [Kautz, 87; Kautz Allen, 86; Allen et al. 91; van Beek Cohen, 91] or syntactic parsing [Vilain, 90] However, in a realistic context plan recognition must be done on an evolving situation described at a lower level, consisting of punctual events. In such a context, the presence of massive, noisy, and ....

Kautz H.A., A Formal Theory of Plan Recognition, PhD Dissertaion, Dept. of Computer Science, University of Rochester, Technical Report 215, 1987.

....Sum From Array Direct Sum Sum Array Fill Array Print Calc Input Sum Figure 3.2: An Example Action Hierarchy deduction are rooted in the exhaustive body of knowledge about actions in a particular domain encoded in the form of an action hierarchy. The hierarchy of Figure 3. 1 taken directly from [Kautz, 1987] depicts specialization relations as dark arrows from specific to general actions. The thinner lines encode decomposition of actions into a set of subactions. Not encoded in this figure are additional domain constraints such as temporal relations among sub actions, although this information is ....

....all of them. When this heuristic is applied, the result is a set of restrictive assertions about the functions of each observed actions. If this causes an inconsistency, the system backtracks up the explanation path to where the simplicity heuristic incorrectly merged the explanation paths. Kautz [Kautz, 1987] identifies two primary problems that must be dealt with in incre CHAPTER 3. PLAN RECOGNITION 58 mental recognition systems. The first of these is the combinatorial problem which arises when the minimum cardinality assumption is relaxed to include two or more primary actions. This relaxation ....

[Article contains additional citation context not shown here]

Henry Kautz. A Formal Theory of Plan Recognition. PhD thesis, University of Rochester, Department of Computer Science, Rochester, New York, 1987. BIBLIOGRAPHY 309

....domain, nor does it depend on the observed agent having a correct plan, from the perspective of the observer. Finally, the argumentation system provides a precise, comprehensible rationale for why one conclusion should be preferred over others. The Circumscriptive Model of Plan Recognition Kautz [43, 44, 45] has proposed an entirely different way of approaching the plan recognition problem, one based on McCarthy s circumscription scheme [55] which minimizes the set of inferences that can be derived from an initial group of assumptions. Every observed action is assumed to be part of some highlevel ....

Henry Kautz. A Formal Theory of Plan Recognition. PhD thesis, Department of Computer Science, University of Rochester, 1987. Technical Report 215.

....the input is supposed to be correct and consists of actions, that is, chunks of activity representing clear cut transitions in the situation that can serve as building bricks to higher plans. This case corresponds to strong theory domains which can receive a formal treatment, such as abduction [Kautz, 87; Kautz Allen, 86; Allen et al. 91; van Beek Cohen, 91] or syntactic parsing [Vilain, 90] although this does not preclude non formal treatment, as shown by the object oriented mechanisms in [Delannoy, 90] However, in a realistic context plan recognition must be done from lower level input, ....

Kautz H., A Formal Theory of Plan Recognition, PhD Dissertaion, Dept. of Computer Science, University of Rochester, Technical Report 215, 1987.

....Some of the more influential formalist work is Discourse Representation 1 With regard to dialogue, research has focused on cooperative plan based endeavors such as tutoring and interactive explanation. As a result, many discourse generation ideas are shared with work on plan recognition [Kautz 87, Hobbs et al. 88, Charniak Shimony 90] Several research efforts are investigating the nature and role of participants beliefs and intentions [Pollack 86, Cohen Levesque 90, Grosz Sidner 90, Lochbaum 91] and much effort is focused on the types of plans that underlie this type of discourse ....

Kautz, H. 1987. A Formal Theory of Plan Recognition. Ph.D. dissertation, University of Rochester.

....of all three types of spatial relations. On the other hand, temporal representation and reasoning is an important recurring research topic mainly in AI community (see for example [32] 3] 16] with several applications like natural language processing [2] planning [14] plan recognition [15], diagnosis [26] knowledgebased systems [17] etc. Undoubtedly, Allen s temporal interval algebra [1] was the most influential work. A few researchers [23, 12] have suggested that this algebra could be extended to represent spatial relationships. Such attempts again have limited applicability as ....

Kautz, H. A., "A Formal Theory of Plan Recognition", PhD Thesis, Technical Report 215, University of Rochester, 1987.

....1985; Hobbs et al. 1988) diagnosis, scientific theory formation and image interpretation. For instance in text interpretation a reader infers the goals and plans of the characters in order to explain the events and actions described in the text (this is known as plan recognition (Allen 1995; Kautz 1987)) In medical diagnosis, based on observed symptoms of a patient, a physician infers possible diseases that may explain the symptoms, while in image interpretation, a vision system infers the objects present in a scene that may explain the image. In this paper we introduce a number of orthogonal ....

Kautz, H. A. 1987. A Formal Theory of Plan Recognition. Ph.D. Dissertation, University of Rochester, Department of Computer Science. Also availiable as TR 215.

....approaches they are treated as opposites. Much of the discussion in this paper has not been specific to diagnosis, but can be applied to any recognition task, where the problem is to determine what is in a system (or a picture) based on observations of the system. For example, one can see [ Kautz, 1987 ] as using the idea of consistency based diagnosis with faults corresponding to plan objects. This paper indicates that there is a corresponding abductive theory of plan recognition. The results of this study indicate that there may not be a representation of knowledge that is independent of how ....

H. A. Kautz, "A Formal Theory for Plan Recognition", Department of Computer Science, University of Rochester, Technical Report TR-215.

....as a plan first requires that one establish a representation of plans as a named or similarly identified set or sequence of actions. With that plan representation and given a set of observed actions, a plan recognition system constructs the set of possible plans which explain the specified actions [7]. Like all recognition tasks, the object to be recognized has to be described in terms of some number of components. Sentences are sequences of words, programs are sequences of instructions, and plans might simply be conceived as named sequences of actions. The concept of sequence is typically ....

....a basis for reasoning in this kind of incomplete information context. At least four different general plan recognition strategies exist in literature: parsing [15] inference under uncertainty [2] plausible inference [1] and circumscribing a hierarchical representation of plans using deduction [7]. The method proposed here uses ideas related to at least the last three of these methods, developed within a framework for managing and reasoning about beliefs. We use the Ghose Goebel belief change model to maintain a dynamic set of assertions that represent actions and relationships amongst ....

[Article contains additional citation context not shown here]

H. Kautz. A Formal Theory of Plan Recognition. PhD thesis, University of Rochester, Department of Computer Science, 1987.

....now propagates down the hierarchy, deriving a more restrictive set of assertions about the top level actions and their subactions. If an inconsistency is detected then the number of top level acts is incremented and the system backtracks to the point at which the simplicity heuristic was applied. Kautz (Kautz 1987) identifies two primary problems that must be dealt with in incremental recognition systems. The first of these is the combinatorial problem which arises when the minimum cardinality assumption is relaxed to include two or more primary actions. This relaxation allows the number of possible ways of ....

....the MakeMarinara action is being shared by MakePastaDish and MakeMeatDish. When such action sharing occurs between plans, additional constraints such as temporal relationships can be very useful in limiting disjunctive conclusions. Kautz s algorithm There are three versions of the Kautz approach(Kautz 1987). Each version is based on a different interpretation about how to integrate or group multiple observation explanations and implement the concept of the minimal explanation. The first, Non dichronic, returns the same result independent of the order of observation of events, and identifies the ....

[Article contains additional citation context not shown here]

Kautz, H. (1987), A Formal Theory of Plan Recognition, PhD thesis, University of Rochester, Department of Computer Science, Rochester, New York.

....of the plan library required for serious applications. Various default reasoning techniques have been applied to the former (e.g. weighted abduction [8] but the latter difficulty has hardly been addressed (but for an exception, see the PHI system [20] Plan recognition is usually performed [109] under the strong assumptions that 1) the recognizer agent has complete knowledge of the domain, and that 2) the agent whose plan is being inferred has a correct plan. These assumptions are clearly not universally true: users may find ways of doing things that the designers of the systems had not ....

....reasoning behavior in closer correspondence with intuition. Numerous formal approaches have been developed in support of making inferences in the absence of complete and reliable information (see, for instance, Brewka [32] Konolige [117] Geffner [81] Reiter [167] Etherington [67] and Kautz [109]) The contribution in this thesis is built upon the Theorist formalism developed by Poole [159] 82 4.1.1 Default Programming with Theorist To make the following presentation more precise, the simple hypothetical reasoning framework of Theorist [162] is used. Vanilla Theorist is defined in ....

Henry A. Kautz. A formal theory of plan recognition. Technical Report 215, Dept. of Comp. Sci., University of Rochester, Rochester, NY 14627, 1987.

....that has been appropriately abstracted through state reduction, the plan recognition component of AC attempts to reason about plans and goals. In this section I will review several plan recognition techniques and discuss their applicability to AC . 3. 3 Formal theory of Plan recognition Kautz [14] presents a framework for reasoning about agent plans which is expressive and comes with formal guarantees. It includes an event hierarchy which is a series of abstractions of core events. Each event in the hierarchy has constraints placed on it and the constraints may be based on other events, ....

H. A. Kautz. A Formal Theory of Plan Recognition. PhD thesis, Univeristy of Rochester, Rochester, NY, May 1987. Technical Report 215.

No context found.

H. A. Kautz. A Formal Theory of Plan Recognition. PhD thesis, University of Rochester, 1987.

No context found.

H. Kautz. A Formal Theory Of Plan Recognition. PhD thesis, University of Rochester, 1987.

No context found.

Henry A. Kautz, \A Formal Theory of Plan Recognition," Technical Report 215, University of Rochester, Department of Computer Science, 1987, PhD thesis.

No context found.

Kautz, Henry, 1987. "A Formal Theory of Plan Recognition", Technical Report 215, Department of Computer Science, University of Rochester, Rochester, New York.

No context found.

Kautz, Henry, 1987. "A Formal Theory of Plan Recognition", Technical Report 215, Department of Computer Science, University of Rochester, Rochester, New York.

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