Martha E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th conference on Association for Computational Linguistics, pages 207--214. Association for Computational Linguistics, 1986.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....Incomplete Erroneous Plans This section could possibly be better labeled when users make mistakes . Most systems described above contain the assumption that both the system and the user have exactly the same knowledge about plans in a domain (and non domain plans as well) Both Pollack [Pol86a, Pol86b, Pol87, Pol90] and Quilici [Qui89] show that this assumption is not wellfounded, especially in situations where the user is a novice, and is interacting with the system (an expert) In these cases, as well as in everyday interaction, users may have either incomplete or incorrect knowledge. ....

Martha E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the Twenty-Fourth Annual Meeting of the Association for Computational Linguistics, pages 207--214, New York, June 1986.

.... is expli citly requested (Allen [1] Luria [16] handle prag matically ill formed queries and resolve some inter sentential ellipses (Carberry [2, 3, 4] provide an explanation from the appropriate perspective (McKeown et al. 17] respona to queries that result from an invalid plan (Pollack [20, 21, 22]) and avoid misleading responses and produce user specific cooperative responses (Joshi et aJ. 10, 11] van Beek and Cohen [26, 27] Cohen et al. 7] Example 1 (Joshi et al. 11] As an example of a cooperative response consider the following exchange between student and student advisor ....

Pollack, M. E. 1986. A Model of Plan Infer- ence that Distinguishes Between the Beliefs of Actors and Observers. Proc. of the Jth Conference of the Association for Computational Linguistics, New York, N.Y., 207-214.

....these facets of real discourse. To model situations in which agents share joint plans, one must first ask what it means for two agents to have a joint plan. Grosz and Sidner [5] are concerned with precisely this question, extending to joint plans Pollack s earlier definition of having a plan [20], which was designed for single agent plans. We paraphrase their definition as follows: Two agents have a shared plan to do action A if and only if for each subaction involved in doing A: 1. They mutually believe (a) That the subaction relates in a particular way to A; b) That one of them can ....

Pollack, M. E. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th Meeting of the Association for Computational Linguistics (June 10--13, New York, N.Y.). Association for Computational Linguistics, N.J., 1986, 207--214.

.... TO USER QUERIES IN A COLLABORATIVE ENVIRONMENT Jennifer Chu Department of Computer and Information Sciences University of Delaware Newark, DE 19716, USA Internet: jchu cis.udel.edu Abstract We propose a plan based approach for responding to user queries in a collaborative environment. We ague that in such an environment, the system should not accept the user s query automatically, but should consider it a proposal open for ....

....dialogue system by 1) extending the model in [6] to eliminate the assumption that the system will automatically answer the user s questions or follow the user s proposals, and 2) capturing the notion This materiM is based upon work supported by the National Science Foundation under Grant No. IRI 9122026. of cooperative responses within an overall collaborative framework that allows for negotiation. 2 The Tripartite Model Lambert and Ca,berry proposed a plan based tripartite model of expert novice consultation dialogue which includes a domain level, a problem solving level, and a ....

[Article contains additional citation context not shown here]

Martha Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proc. ACL, pages 207--214, 1986.

....beliefs as cribed to the user during the course of the conversation and how strongly each belief is held. Researchers [1, 3, 5] have noted that discourse understanding can be enhanced by recognizing a user s goals, and that this recognition process requires reasoning about the agent s beliefs [7]. For example, in order to recognize from utterance IS2 in the following dialogue that the speaker has the communicative goal of expressing surprise at the proposition that Dr. Smith is teaching CIS360 and not just asking if Dr. Smith is teaching CIS420, it is necessary for the system to be able ....

....may contribute to, or link to, an action on an immediately higher level. For example, discourse actions may be executed to attain the knowledge needed for problem solving actions at the middle level. This tripartite, plan based model of discourse fa 2The DM is really a mental model of intentions [7] which im pllcitly captures a number of intentions that are attributed to the participants, such as the intention that the participants follow through with the subactions that are part of plans for actions in the DM. 349 cilitates recognition of changing beliefs as the dialogue progresses. ....

Martha Pollack. A model of plan inference that distin- guishes between the beliefs of actors and observers. In Proceedings of the Jth Annual Meeting of the Association for Computational Linguistics, pages 207-214, New York, New York, 1986. 350

....Incomplete Erroneous Plans This section could possibly be better labeled when users make mistakes . Most systems described above contain the assumption that both the system and the user have exactly the same knowledge about plans in a domain (and non domain plans as well) Both Pollack [Pol86a, Pol86b, Pol87, Pol90] and Quilici [Qui89] show that this assumption is not wellfounded, especially in situations where the user is a novice, and is interacting with the system (an expert) In these cases, as well as in everyday interaction, users may have either incomplete or incorrect knowledge. ....

Martha E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the Twenty-Fourth Annual Meeting of the Association for Computational Linguistics, pages 207-214, New York, June 1986.

....the error) This algorithm is actually very similar to what Schmidt et al. Schmidt et al. 1978] also Section 2.3 above) report that humans did in experiments. Humans would come to just one conclusion and then revise it if later information invalidated it. Both Pollack ( Pollack, 1986a; Pollack, 1986b; Pollack, 1987; Pollack, 1990] and Quilici ( Quilici, 1989] have shown that the assumption that the user is always correct is a limiting one. It may be possible to use a heuristic (based on how far the system has to backtrack, etc. to determine if the system should query to make sure the user ....

Martha E. Pollack, "A Model of Plan Inference that Distinguishes between the Beliefs of Actors and Observers," In Proceedings of the Twenty-Fourth Annual Meeting of the Association for Computational Linguistics, pages 207--214, New York, June 1986.

....B is an assertion related to another subgoal on the theorem proving tree as shown in Figure 1. The user may initiate such a change in subdialogue in an attempt to pursue another path to the global goal. Here the machine first must track the user s intention (in a process called plan recognition [15, 16, 17, 18, 19]) and then evaluate whether to follow the move or not. This decision is based upon the current level of the initiative of the system as described below. If the system follows the user s initiative, it will apply its internal theorem proving system to the subgoal E and pursue voice interactions ....

M.E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th Annual Meeting of the Association for Computational Linguistics, pages 207--214, 1986.

....facets of real discourse. If one wishes to model situations in which agents share joint plans, one must ask what it means for two agents to have a joint plan to do something. Grosz and Sidner [5] are concerned with precisely this question. They extend Pollack s earlier definition of having a plan [20], which was designed only for single agent plans, to the case of joint plans. We paraphrase their definition as follows: Definition: Two agents have a shared plan to do action A if and only if for each subaction involved in doing A: 1) they mutually believe (a) that the subaction relates in a ....

Pollack, Martha E. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th meeting of the Association for Computational Linguistics (June 10--13, New York, N.Y.). Association for Computational Linguistics, N.J., 1986, pp. 207--214.

.... participants have to perform: volunteering the information that is helpful to achieve the goal of the dialogue partner even if the partner does not ask for the information [ Allen and Perrault, 1980 ] pointing out that the partner has invalid intentions or false beliefs [ Carberry, 1986; Pollack, 1986] and negotiating with the partner to resolve the conflict in the participants beliefs [ ChuCarroll and Carberry, 1995 ] These collaborative tasks are important, but this paper focuses on the fundamental collaborative task of maintaining the mutual understanding of exchanges by handling the ....

Martha E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of 24th Annual Meeting of the Association for Computational Linguistics, pages 207--214, 1986.

....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 (see [Litman 85, Lambert Carberry 91, Ramshaw 91] Most of these theories postulate several levels of plans, each level handling a distinct phenomenon ....

.... SPOKESMAN [Meteer 90] PIT [Kreyss Novak 90] POPEL [Reithinger 91] JOYCE [Rambow Korelsky 92] and others) and taking as far as possible into account the theoretical work of [Grosz Sidner 86, Asher 92, 11 Polanyi 88] and the work on intention recognition [Allen Perrault 80, Litman 85, Pollack 86, Lambert Carberry 91] the following general assertions about the structure of plan based English discourse can be formulated: 1. Discourse: A discourse (a text) is a structured collected of clauses. By their semantic relatedness, clauses are grouped into segments; the discourse structure is ....

Pollack, M.E. 1986. A Model of Plan Inference that Distinguishes between the Beliefs of Actors and Observers. Proceedings of the 24th Annual Conference of the Association of Computational Linguistics, Chicago, (207--214).

....belief mutually known, thus allowing an agent to abandon his commitment to the joint goal with no further action towards the remaining team members. SharedPlans The SharedPlan model of collaboration [10] was developed to account for several deficiencies noted in Pollack s mental state of plans [21, 22], these being spelled out in [9] 1) an action being performed by two or more agents could be decomposed into actions being performed by each individual agent. Grosz [7] provides a lucid discussion of why a joint action is not the sum of the individual plans, and hence why collaboration cannot be ....

Pollack, M. "A Model of Plan Inference that Distinguishes Between the Beliefs of Actors and Observers", Proc. 24th Ann. Mtg. Assoc. for Computational Linguistics, (1986), 207-214.

....can take the form of actions that are intended to advance the agent s goals but which are inappropriate[Pol87] Most plan inference systems assume that the user s knowledge is a subset of the system s and fail to account for the presence of misconceptions about how to achieve domain goals. Pollack[Pol86] was the first to explicitly ascribe beliefs to an agent during plan inference. Although her system could ascribe the system s beliefs about recipes for achieving goals, it could also ascribe principled variations of those recipes. This allowed Pollack to infer ill formed plans and thus account ....

....with very simple variations, such as the omission of a constraint; extending her work to more complex variations requires extensive research. CalistriYeh [CY91] presents a classification of plan based misconceptions that encompasses most of the categories identified by other researchers[Pol86, QDF88, vB87]. He presents a plan inference system that uses estimated probabilities to guide the search for an agent s intentions. Formulas attached to each class of misconception are used to estimate the probability of a particular kind of misconception based on the plan under consideration. Application of ....

Martha Pollack. A Model of Plan Inference that Distinguishes Between the Beliefs of Actors and Observers. In Proceedings of the 24th Annual Meeting of the Association for Computational Linguistics, pages 207--214, New York, New York, 1986.

....This is based on Lambert and Carberry s plan recognition algorithm [Lambert and Carberry, 1991] augmented to allow the system to ascribe to the executing agent erroneous beliefs that it hypothesizes the executing agent may reasonably hold. This builds on research on recognizing ill formed plans [Pollack, 1986] and work on a relaxation algorithm for flexible plan recognition in dialogue [Eller and Carberry, 1992] The proposal evaluation and modification processes discussed in this paper operate on the dialogue model that would be generated by such a hypothesized recognition algorithm given a set of ....

....agents [Sidner, 1994] The evaluation of a proposal is carried out in two parts, the evaluation of proposed actions and the evaluation of proposed beliefs. First, the agent must decide whether or not she believes that the proposed domain and problem solving actions will contribute to a valid plan [Pollack, 1986] as well as produce a reasonably efficient way to achieve the agents high level goal [Joshi et al. 1984, van Beek, 1987] The validity of a set of proposed actions is evaluated based on an agent s beliefs about recipes for performing actions, as well as her beliefs about the characteristics of ....

[Article contains additional citation context not shown here]

Martha E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th Annual Meeting of the Association for Computational Linguistics, pages 207--214, 1986.

....a proposal without the other agent s consent; thus, before the actual modification of the proposal, the conflict between the agents must have been resolved. This is captured by the precondition of Modify Node, the first subaction of Correct Node. The attempt to satisfy this precondi 4 A recipe [21] is a template for performing an action. It contains the applicability conditions for performing the action, the subactions comprising the body of the action, etc. 5 Applicability conditions are conditions that must already be satisfied in order for an action to be reasonable to pursue, whereas ....

Martha E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the ACL, pages 207--214, 1986.

....have been expanded. If so, then check if the plan achieves G. If so, then go back to step 2. If not, then remove the current action choice and check for others. Figure 3: A NOAH style planning algorithm and intentions of the agents involved in the planning and how the agents are cooperating [53, 54, 26]. Action operators also do not allow one to represent the different kinds of relationships between subactions and effects, maintenance goals, or simultaneous actions. A significant amount of information that is needed to handle plan failure, modification and explanation is therefore missing. ....

....planning representation cannot adequately model collaboration. Because of this, more recent work in this area has tended to fall in one of three areas: redefining the term plan ; hypothesizing different levels of plans; and redefining the planning process. 4. 1 Redefining the term plan Pollack [53, 54] distinguishes between two views of planning; that in which plans are viewed as data structures or recipes for action , and that in which a plan is a structured collection of beliefs and intentions . The first approach uses a set of action operators and assumes that the beliefs of all agents ....

M. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the Meeting of the Association for Computational Linguistics, New York, NY, 1986.

....needed that takes into account the syntactic form of utterances in modeling the strength of different user beliefs. A number of researchers have investigated techniques for inferring a model of an agent s plan [Allen and Perrault1980, Carberry1988, Litman and Allen1987, Lambert and Carberry1991, Pollack1986] Allen formulated heuristics for identifying obstacles in the inferred domain plan, including knowledge deficits that could be overcome via helpful information from the respondent. In addition, Allen [Perrault and Allen1980] showed how plan recognition techniques could be used to recognize the ....

Pollack, M.: 1986, `A Model of Plan Inference that Distinguishes Between the Beliefs of Actors and Observers'. In: Proceedings of the 24th Annual Meeting of the Association for Computational Linguistics. New York, New York, pp. 207--214.

....are one of the primary mechanisms needed for tracking the conversation as it jumps from subdialog to subdialog. This is known elsewhere as the plan recognition problem and it has received much attention in recent years. See, for example, Allen, 1983 ] Litman and Allen, 1987 ] Pollack, 1986 ] and [ Carberry, 1990 ] Systems capable of all of the above behaviors are rare as has been observed by [ Allen et al. 1989 ] no one knows how to fit all of the pieces together. One of the contributions of the current work is to present an architecture that can provide them all in the ....

M.E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th Annual Meeting of the Association for Computational Linguistics, pages 207--214, 1986.

....involves matching incoming utterances to expected interactions for the various available subgoals and following the user to subgoals where matches are found. This is called plan recognition in the literature and has been the object of much study (Allen and Perrault, 1980; Litman and Allen, 1987; Pollack, 1986; Carberry, 1988; Carberry, 1990) Mechanisms for both managing movement between subgoals and deciding when to release control to the other participant, including extensive analyses of their effectiveness, are given in (Smith and Hipp, 1994; Guinn, 1995; Guinn, 1996) 1.3 Accounting for the user ....

M. E. Pollack. 1986. A model of plan inference that distinguishes between the beliefs of factors and observers.

....to a system, or evaluating and improving one that is already in place. A discourse module might combine theories on, e.g. centering or local focusing [GJW83, Sid79] global focus [Gro77] coherence relations[Hob85] event reference [Web86] intonational structure [PH90] system vs. user beliefs [Pol86], plan or intent recognition or production [Coh78, AP80, SI81] control[WS88] or complex syntactic structures [Pri85] How might one evaluate the relative contributions of each of these factors or compare two approaches to the same problem In order to take steps towards establishing a ....

Martha Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proc. 24th Annual Meeting of the Association of Computational Linguistics, pages 207-- 214, Columbia University, N.Y., N.Y, 1986.

....problem is well known in literature on plan recognition, and it is unlikely to ever get a satisfactory solution for keyhole plan recognition. There exists solutions for intended plan recognition that can deal with recognizing instances of novel plans or user specific preferences or misconceptions [Pollack 1986], CalistriYeh 1991] However, the only possible approach to achieving any of this in keyhole plan recognition is to introduce some mechanism of learning over time based on probabilistic data, and it is unclear if such training can be achieved that can recognize completely novel plans. Bauer ....

M. E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. 1986. in Proc. of the 24th annual meeting of the assoc.

....as whether or not the proposed beliefs conflict with her existing beliefs. The following sections briefly describe the processes for detecting conflicts in proposed actions and beliefs. For further details on the conflict detection processes, see [3, 6] 3. 1 Evaluating Proposed Actions Pollack [27] argued that a plan can be invalid because one of its actions is infeasible, meaning that the action cannot be performed by its agent, or because the plan itself is ill formed, which occurs when a child action in the plan does not contribute to its parent action as intended. Furthermore, Joshi et ....

....4 The process starts at the leaf nodes of the proposed 2 In the air traffic control domain, these preferences can be viewed as qualities of a resultant plan that are generally preferred by most agents, such as preferring re routing flights to the nearest possible destination. 3 A recipe [27] is a template for performing an action. It encodes the applicability conditions and preconditions for an action, the effects of an action, the subactions comprising the body of an action, etc. 4 In this paper, we make the assumption that an agent can always come to a decision regarding the ....

Martha E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th Annual Meeting of the Association for Computational Linguistics, pages 207--214, 1986.

....language interface must be able to participate in a coherent dialogue with the user. A common, generally applicable approach is to use plan inference as a basis for reasoning about intentions of the user as proposed by, for instance, Allen Perrault (1980) Litman (1986) Carberry (1989) and Pollack (1986). However, computationally these approaches are not so efficient. Reichman (1985) describes a discourse grammar based on the assumption that a conversation can be described using conventionalized discourse rules. Gilbert, Buckland, Frolich, Jirotka Luff (1990) uses interaction rules in their ....

Pollack, Martha E. (1986) A Model of Plan Inference that Distinguishes between the Beliefs of Actors and Observers, Proceedings of the 24th Annual Meeting of the ACL, New York.

....and clear, at least in theory. However, they suffer from numerous disadvantages. A plan representation for discourse should encode not just actions and their preconditions and effects, but also the beliefs and intentions of the agents involved in the planning and how the agents are cooperating [82, 83, 35]. Action operators also do not allow one to represent the different kinds of relationships between sub actions or effects, maintenance goals, or simultaneous actions. A significant amount of information that is needed to handle plan failure, modification and explanation is therefore missing. ....

....plan failure, modification and explanation is therefore missing. Because of this, more recent work in this area has tended to fall in one of three areas: redefining the term plan ; hypothesizing different levels of plans; and redefining the planning process. Redefining the term plan Pollack [82, 83] distinguishes between two views of planning; that in which plans are viewed as data structures or recipes for action , and that in which a plan is a structured collection of beliefs and intentions . The first approach uses a set of action operators and assumes that the beliefs of all agents ....

[Article contains additional citation context not shown here]

M. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the Meeting of the Association for Computational Linguistics, New York, NY, 1986.

....the evaluator performs a top down analysis to detect invalid as well as suboptimal plans. A plan is considered invalid if either an action in the plan cannot be performed (an action is infeasible) or a child action does not contribute to its parent action as intended (the plan is ill formed (Pollack 1986)) A plan is considered suboptimal if there exists a better way to perform the desired action. We evaluate the optimality of a plan with respect to EA s preferences (Elzer, Chu Carroll, Carberry 1994) The evaluation of actions is a top down process which terminates as soon as a conflict ....

....is teaching AI next semester, and 3) Dr. Smith teaching AI provides support for the belief that he is not going on sabbatical. The process for evaluating proposed beliefs starts at the leaf node of the proposed belief tree, Teaches (Smith,AI,next semester) The system searches for 4 A recipe (Pollack 1986) is a template for performing actions. It contains the preconditions of an action, the subactions comprising the body of an action, etc. 5 Applicability conditions are conditions that must already be satisfied in order for an action to be reasonable to pursue, whereas an agent can try to achieve ....

Pollack, M. E. 1986. A model of plan inference that distinguishes between the beliefs of actors and observers.

....continue to talk at cross purposes. Alternatively, the conversation might break down, leading one participant or the other to 1 Misunderstanding should not be confused with misconception. A misconception is an error in the prior knowledge of an participant. McCoy [26] Calistri Yeh [5] Pollack [33, 34], and others have studied the problem of how one participant can, in conversation, determine the misconceptions of another in order to correct them. determine that a misunderstanding has occurred. It is thus useful to divide misunderstanding into two types: self misunderstandings are those that ....

Pollack, Martha E. "A model of plan inference that distinguishes between the beliefs of actors and observers." Proceedings of the 24th annual meeting of the Association for Computational Linguistics, New York, 1986, 207--214.

....(5) to the correct invalid proposal action specified by utterances (6) and (7) or initiates a new topic that forms the root node of a separate discourse tree. A discourse understanding system capable of performing such a task typically contains three components: 1) a library of generic recipes (Pollack, 1986), 2) the plan inference module, and 3) the system s private knowledge about the domain and about the world, and (optionally) its beliefs about the dialogue participants and their beliefs. Below we sketch the outline of the discourse understanding process. Header: ....

....actions, and results in a chain of hypothesized actions. A parent action (A i ) may be chained to an inferred or observed action (A j ) if one of two conditions holds. First, A j may be in the body of the recipe for performing A i ; in other words, performing A j contributes to performing A i (Pollack, 1986). In this case, we may hypothesize that A j is performed as part of carrying out the higher level action A i . Second, the goal of A j may match a precondition in A i ; in other words, performing A j enables performing A i (Pollack, 1986) In this case, we may hypothesize that A j is performed in ....

[Article contains additional citation context not shown here]

Pollack, Martha E. 1986. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th Annual Meeting of the Association for Computational Linguistics, pages 207--214.

....domain plan being constructed, or when they 2 A recipe is a template for performing an action. It encodes information such as the preconditions for executing the action, the effects of the action, the subactions comprising the body of the action, etc. We adopt the term recipe following Pollack [Pol86]. Build Plan(U, S, Learn Material(U,CS180, Dr.Smith) Enable aarc Subaction arc Key: Learn Material(U, CS180, Dr.Smith) Build Plan(U, S, Take Course(U,CS180) Problem Solving Level Domain Level Obtain Info Ref(U, S, time, Meets(CS180, Dr.Smith, time) Ask Ref(U, S, time, Meets(CS180, ....

....Gal92] Although this system models negotiations that occur when agents have inconsistent beliefs, it does not recognize conflicts unless they are directly stated in an utterance,nor does it provide general strategies for responding to a detected conflict. In addition, several other researchers[JWW84, Pol86, vB87] have addressed generating cooperative responses and responding to plan based misconceptions, but did not capture these within an overall collaborative system that must negotiate proposals with the user. Our plan based approach, on the other hand, provides a general mechanism for reasoning about ....

Martha Pollack. A Model of Plan Inference that Distinguishes Between the Beliefs of Actors and Observers. In Proceedings of the 24th Annual Meeting of the Associationfor Computational Linguistics, pages 207--214, New York, New York, 1986.

....are plans only in a very limited sense, and that the inference done is of a very simplistic kind. Plan Debugging One important use of the inferred plans is to debug them, that is, recognize when a user has an faulty or suboptimal way to fulfil a certain goal. This task has been tackled by Pollack (1986). Although interesting, Pollack s work has some shortcomings which makes it insufficient for the intelligent help domain. Firstly, with Pollack s approach no real plan inference is really taking place. In fact, she assumes that the whole plan is stated by the user in the questions posed to the ....

Pollack, Martha, A Model of Plan Inference that Distinguishes between the Beliefs of Actors and Observers, in Proc. of the 24th Annual Meeting of ACL, 1986.

....avoid assumptions about the correctness or completeness of the agents beliefs. Early work on plan recognition in discourse, e.g. Allen Perrault (1980) Sidner Israel (1981) was based on work in AI planning systems, in particular the STRIPS formalism (Fikes and Nilsson, 1971) However, as Pollack (1986) has argued, because these systems do not differentiate between the beliefs and intentions of the different conversational participants, they are insufficient for modelling discourse. Although Pollack proposes a model that does make this distinction, her model has other shortcomings. In ....

.... e.g. sequence(boil(water) add(noodles,water) represents the sequence of doing an act of type boil(water) followed by an act of type add(noodles,water) while CGEN(mix(sauce,noodles) make(pasta dish) C) represents that the first act type conditionally generates the second (Goldman, 1970; Pollack, 1986) . Table 1 lists the act type and corresponding activity relations and constructors that will be used in this paper. Act type constructors and relations are used in specifying recipes. Following Pollack (1990) we use the term recipe to refer to what an agent knows when the agent knows a way of ....

Pollack, M. E. 1986. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th Annual Meeting of the ACL, pages 207--214, New York, NY.

....discrepancy between plan recognition and plan evaluation also disappears. For plan evaluation, both the intended goal and the observed actions can be treated as conditions on a proof. Plan evaluation is useful in particular when the user may have misconceptions about the domain. Following Pollack [13], it seems useful to seek a proof for a predicate that finds differences between the beliefs of the acting agent and those of the observer. Since the applications of the axiom rule signify how we have used the observations as conditions on a proof, it will in some domains be useful to save the ....

Martha E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. in Proc. of the 24th annual meeting of the assoc. for Computational Linguistics, June 1986.

....do not distinguish between the speaker s and hearer s beliefs about actions. Pollack, in modelling how an inferring agent reasons about another agent s invalid plans, differentiates each agent s individual beliefs and intentions regarding actions and the relations among them from other agents (Pollack, 1986; Pollack, 1990) Her model thus provides a useful base on which to define a model of collaborative activity. However, she also makes the assumptions that the inferring agent has complete and accurate knowledge of domain actions and that that agent is recognizing the plans of another. In ....

....and describe two new constructs: recipes and the Contributes relation. We then present an augmentation algorithm and give examples of its use. Finally, we describe future directions of this work. Definition of SharedPlan The definition of SharedPlan follows Pollack s work on single agent plans (Pollack, 1986; Pollack, 1990) in taking the notion of having a plan to be central and to consist of being in a certain mental state, namely, holding certain beliefs and intentions regarding acts and their executability. This stance differs from other work in planning which takes a plan to be a data structure ....

[Article contains additional citation context not shown here]

Pollack, M. E. June 1986. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th Annual Meeting of the ACL.

....it, and if not, attempts to modify the original proposal to a form that will be accepted by both agents. The arbitrator has two components, the evaluator and the modifier. 2. 1 The Evaluator The evaluator checks for two types of discrepancies in beliefs: one that causes the proposal to be invalid [Pollack, 1986] , and one in which the proposal is considered suboptimal [Joshi et al. 1984, van Beek, 1987] A plan is considered invalid if it contains either an infeasible action or a pair of parent child actions in which the child action does not contribute to the parent action as intended. The optimality ....

Martha E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th Annual Meeting of the Association for Computational Linguistics, pages 207--214, New York, 1986.

....discourses that remain if either is removed from the dialogue. Third, although the process of plan construction provides an important context for interpreting utterances, trying to formalize this mental activity under a datastructure approach results in a model that conflates recipes and plans (Pollack, 1990). For example, each of Lambert and Carberry s domain act operators requires as a precondition that the agent have a plan to use that operator to perform the act. That requirement, however, results in the paradoxical situation whereby a recipe for an act ff requires having a plan for ff that uses ....

Pollack, M. E. 1986b. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th Annual Meeting of the ACL, pages 207--214, New York, NY.

....the other s and avoid assumptions about the correctness or completeness of the agents beliefs. Early work on plan recognition in discourse, e.g. Allen Perrault[AP80] Sidner Israel[SI81] was based on work in AI planning systems, in particular the STRIPS formalism[FN71] However, as Pollack[Pol86] has argued, because these systems do not differentiate between the beliefs and intentions of the different conversational participants, they are insufficient for modelling discourse. Although Pollack proposes a model that does make this distinction, her model has other shortcomings. In ....

.... and relations; e.g. sequence(boil(water) add(noodles,water) represents the sequence of doing an act of type boil(water) followed by an act of type add(noodles,water) while CGEN(mix(sauce,noodles) make(pasta dish) C) represents that the first act type conditionally generates the second[Gol70, Pol86]. Table 1 lists the act type and corresponding activity relations and constructors that will be used in this paper. Act type Activity Relations CGEN(fl 1 ,fl 2 ,C) GEN( Gamma 1 , Gamma 2 ) CENABLES(fl 1 ,fl 2 ,C) ENABLES( Gamma 1 , Gamma 2 ) Constructors sequence(fl 1 , fl n ) simult(fl 1 ....

M. E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th Annual Meeting of the ACL, pages 207--214, New York, NY, 1986.

....is made by the evaluator, which will be discussed in this section. This paper only considers cases in which the user s proposal contains an infeasible action (one that cannot be performed) or would result in an ill formed plan (one whose actions do not contribute to one another as intended)[9]. We argue that the evaluator, in order to check for erroneous plans goals, only needs to examine actions in the proposed model, since actions in the existing model would have been checked when they were proposed. When a chain of actions is proposed, the evaluator starts examining from the ....

....as expected. Thus their meta plans do not handle correction of proposed additions to the dialogue model (since this generally does not involve adding a step to the proposal) Furthermore, they were only concerned with understanding utterances, not with generating appropriate responses. The work in [5, 11, 9] addressed generating cooperative responses and responding to plan based misconceptions, but did not capture these within an overall collaborative system that must negotiate proposals with the user. Heeman [4] used meta plans to account for collaboration on referring expressions. We have addressed ....

Martha Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proc. ACL, pages 207--214, 1986.

....to generalize their expectations beyond this domain. Another group of researchers that have used expectation are those who have worked in the field of plan recognition. Examples of this work can be found in Allen and Perrault [6] Allen [2] Litman and Allen [70] Sidner and Israel [108] Pollack [91], Carberry [18] Ramshaw [98] and [99] and Lambert and Carberry [67] Their use of expectation consists of the descriptions of plans which speakers may be attempting to complete. It is expected that a speaker is attempting to complete one of the specified plans. Plan recognition entails ....

....in the dialog should mirror the level of initiative in the task. Furthermore, the level of initiative in the task should also indicate the relative priority of each participant s goals. As has been noted by researchers who have worked on the problem of correcting user misconceptions (e.g. Pollack [91], McCoy [75] and Quilici [96] the user s goals and beliefs may differ from the computer s goals and beliefs. However in general, the user s differing goals may not be inappropriate. Thus, the level of task and dialog initiative will also indicate the relative priority of each participant s ....

M.E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th Annual Meeting of the Association for Computational Linguistics, pages 207--214, 1986.

....be other uses for the observations in subsequent proofs than the one utilized in the current proof. The discrepancy between plan recognition and plan evaluation also disappears. For plan evaluation, both the intended goal and the observed actions can be treated as conditionson a proof. Following Pollack (Pollack 1986), it seems useful to seek a proof for a predicate that finds differences between the beliefs of the acting agent and those of the observer. Mechanisms for selecting the most appropriate plan Although the route guidance example shows that the PIDabductive framework can generate such proofs that ....

Pollack, M. E. 1986. A model of plan inference that distinguishes between the beliefs of actors and observers.

....or declarative. Finally, Kautz use of e graphs is computationally very expensive and declaratively undesirable. We believe indeed that the decision to save e graphs to accommodate future observations is counterintuitive. The plan recognition systems of the second generation (see for instance [Pol86, Pol90] have been concerned with robustness, and, hence, the relaxation of some of the assumptions inherent in much of the previous work, e.g. that the agents construct plans that are correct. By its recognition of inconsistent plans, our model does indeed find those cases where the ....

....the observations. As of now, however, our model does not make use of an attitudes model (beliefs, intentions, etc. In this paper, we have not assumed that the recogniser has any prior knowledge as to the goals of the user, unlike for instance the work of Helft and Konolige [HK90] or Pollack [Pol86, Pol90] We finally point out that one could possibly think of an alternative implementation of our algorithms using Assumption based Truth Maintenance Systems [Doy79, de 86] It seems indeed that a two layer ATMS could be used to implement the abductive reasoning on the Concept KB, and the ....

M. E. Pollack. A Model of Plan Inference that distinguishes between the Beliefs of Actors and Observers. In P. Georgeff, M and A. Lansky, editors, the 1986 Workshop: Reasoning about Actions and Plans, pages 279--295, 1986.

....class of cues, analytical cues, are cues that cannot be recognized unless the hearer performs an evaluation on the speaker s proposal using the hearer s private domain knowledge. After the evaluation, the hearer may find the speaker s proposal 1) invalid, if an action in the proposal is infeasible (Pollack 1986; Chu Carroll Carberry 1994) if the proposed plan is ill formed (Pollack 1986; Chu Carroll Carberry 1994) if the hearer disagrees with a proposed mutual belief (Chu Carroll Carberry 1995) or if the speaker s utterances indicate that a miscommunication has occurred (McRoy Hirst 1995; ....

....performs an evaluation on the speaker s proposal using the hearer s private domain knowledge. After the evaluation, the hearer may find the speaker s proposal 1) invalid, if an action in the proposal is infeasible (Pollack 1986; Chu Carroll Carberry 1994) if the proposed plan is ill formed (Pollack 1986; Chu Carroll Carberry 1994) if the hearer disagrees with a proposed mutual belief (Chu Carroll Carberry 1995) or if the speaker s utterances indicate that a miscommunication has occurred (McRoy Hirst 1995; Traum Dillenbourg 1996) 2) suboptimal, if the hearer detects a better ....

Pollack, M. E. 1986. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th Annual Meeting of the Association for Computational Linguistics, 207--214.

....Lambert and Carberry (1991) introduced problem solving plans to represent means of constructing domain plans, while Ramshaw (1991) introduced exploration plans to make reference to domain plans an agent is only considering adopting. These approaches are all based on a datastructure view of plans (Pollack, 1986). Our approach results from recognizing that agents engage in dialogues and subdialogues for reasons that derive from mental state requirements of action and collaboration. These reasons include exploring courses of action; constructing, selecting, and elaborating recipes; deciding upon particular ....

M. E. Pollack. 1986. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th Annual Meeting of the ACL, pages 207--214, New York, NY.

....is that she performed that reasoning. An agent that did not attempt to understand my utterance as part of my plans would have gone ahead and made the car reservation for Tuesday. In my thesis research, I focused on with the problem of recognizing invalid plans, like the one I have just described [51, 52, 54]. I showed that the traditional models of plan recognition, which equate plans with recipes for action, are insufficient for this task. Traditional models are more or less restricted to searching through a fixed space of plans. With some limited exceptions, you can only recognize plans that you ....

M. E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th Annual Meeting of the Association for Computational Linguistics, pages 207-- 214, New York, 1986.

No context found.

Martha E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24th conference on Association for Computational Linguistics, pages 207--214. Association for Computational Linguistics, 1986.

No context found.

M.E. Pollack. A Model of Plan Inference that Distinguishes Between the Beliefs of Actors and Observers. In Proceedings of the 1986.

No context found.

Martha E. Pollack, \A Model of Plan Inference that Distinguishes between the Beliefs of Actors and Observers," In Proceedings of the Twenty-Fourth Annual Meeting of the Association for Computational Linguistics, pages 207-214, New York, June 1986.

No context found.

Pollack, Martha: "A Model of Plan Inference that Distinguishes between the Beliefs of Actors and Observers", in: "Proceedings of the Twenty-fourth Annual Meeting of the Association for Computational Linguistics", New York, 1986, pp. 207-214.

No context found.

Martha E. Pollack. A model of plan inference that distinguishes between the beliefs of actors and observers. In Proceedings of the 24rd Annual Conference of the Association for Computational Linguistics, pages 207--214, 1986.

No context found.

Martha E. Pollack (1986), "A model of plan inference that distinguishes between the beliefs of actors and observers", Proceedings of the 24th annual meeting of the Association for Computational Linguistics, New York, pp. 207--214.

No context found.

Pollack, Martha E. (1986). "A model of plan inference that distinguishes between the beliefs of actors and observers." Proceedings of the 24th annual meeting of the Association for Computational Linguistics, New York, 207--214.

No context found.

Martha E. Pollack, "A model of plan inference that distinguishes between the beliefs of actors and observers," in Proceedings of the Twenty-Fourth Annual Meeting of the Association for Computational Linguistics, pages 207--214, New York, NY, 10--13 June 1986. Columbia University.

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