Allen Newell. The knowledge level. Artificial Intelligence, 18:87--127, 1982.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....function of the system is to be achieved and contains the task decomposition. The model of cooperation assigns tasks and sub tasks to agents. The model of expertise specifies the problem solving expertise required to perform the problem solving tasks assigned to the system at the knowledge level [Newell, 1982]. The conceptual model is an abstract description of the objects and operations the kbs should know about. Finally, the design model is a highlevel specification of the kbs, the operationalisation of which should be the kbs itself. The model that we are most interested in here, as it appears to ....

Allen Newell. The knowledge level. Artificial Intelligence, 18(1):87--127, January 1982.

....are also a subset of an actor s competencies. They describe what an actor is expected to do, e.g. due to a commitment in consequence of a complex negotiation procedure or because of an actor s intrinsic role. The first three criteria correspond to the knowledge level descriptions proposed by Newell [Newell, 1982] . The latter two reflect that the various actors in distributed OMs form a society, not just an accumulation. Rights and obligations are the basis for coordinating the negotiation processes that are needed to create a shared understanding. Figure 3 shows a taxonomy of possible roles which ....

Allen Newell. The knowledge level. Artificial Intelligence, 18(1):87--127, 1982.

....in 1988 the HL used certainty factors, but in 1989 switchedtoafive valued default logic. The Constraint Language has had a two valued logic throughout [Guha, 1990] Although the epistemological heuristic distinction was proposed twenty years ago, and similar proposals have also been made [Newell, 1982, Levesque, 1 In non monotonic reasoning systems unsoundness usually goes hand in hand with incompleteness, because if you fail to derive an exceptional fact (Tweety is a penguin) you will incorrectly derive facts using the assertion it would haveoverridden (Tweety flies) 1 Figure 1:Schematic ....

Allen Newell. The knowledge level. Artificial Intelligence, 18(1):87--127, 1982.

....defined in terms of the initially, happens, initiates, terminates and succeeds predicates, involving both the domain specification and the semantics. 7. 6 On the Knowledge Level and its performances The ego , according to Sandewall s definition, corresponds to Newell s Knowledge Level [New82] of an intelligent agent, and is intended to be that part of the agent where reasoning is performed. Previous sections showed, without loss of generality, how the ego can be identified with its own knowledge base, consisting in a finite number of theories, object level and meta level ones. As an ....

.... Sterling s [Ste84] Expert System = Knowledge Meta interpreter , we here obtained the equation Knowledge Level = Object Level Meta Level Our construction of a calculus for computing feature values at different points in time, perfectly fits with Newell s understanding of the knowledge level [New82] Logic is just a representation of knowledge. It is not the knowledge itself, but a structure at the symbol level. page 110) Logic is the appropriate tool for analyzing the knowledge level, though it is often not the preferred representation to use for a given domain. page 121) To ....

Allen Newell. The Knowledge Level. Artificial Intelligence, 18(1):87-- 127, 1982.

.... of AI research, admittedly, only describes systems of the class identified by the physical symbol system hypothesis: namely, the class of systems which embodies the essential nature of symbols and which is the necessary and sufficient condition for a generally intelligent agent [New82, page 94] Non symbolic approaches to AI for example, those which are embodied in connectionist or artificial neural networks and in genetic algorithms have been omitted from the present discussion but can be read about elsewhere [RK91, pages 483 528] Intelligent Agent Design j 10 path ....

....mental state and consequent action sequences: in other words, by taking to itself the role of the other (i.e. the agent) assuming the agent s goals and attending to the common external environment, such that] the actions it determines for itself will be those that the agent should take. New82, page 109] The purpose of a TouringMachine making these projections, is to determine, in the light of the present state of the world, whether any agent and, indeed, whether the TouringMachine itself is destined for some future conflict with any of the other entities in the world. The ....

[Article contains additional citation context not shown here]

Allen Newell. The knowledge level. Artificial Intelligence, 18:87-- 127, 1982.

....or the metatool implements the knowledgeacquisition tool. Metatools transform the tool specification developed in step 2 to an operational program. Developers can model domains independent of the problem solving method through specification approaches, such as knowledge level analysis [ Newell, 1982 ] 1 Dots is an acronym for Domain Oriented Tool Support. and ontological analysis [ Alexander et al. 1987 ] However, in the general case, the transformation from such models into a high quality knowledge acquisition tool is nontrivial. Target knowledge acquisition tools must be designed by ....

Allen Newell. The knowledge level. Artificial Intelligence, 18(1):87--127, 1982.

....data such as consecutive serological findings. Among those that do is TrenDx [22] utilizing the temporal utility package TUP [6] Instead, authors prefer to focus on deriving temporal abstractions holding over intervals [8, 12, 23] and applying qualitative time measures and relations [7, 9, 24]. However, this approach is not only unsuited for the given problem (extremely few samples leave nothing to abstract) it also suffers from the mismatch of bottom up temporal abstraction and requirements of subsequent symbolic reasoning 1 . Solutions regarding the quantitative nature of the ....

Allen JF: Towards a general theory of action and time. Artificial Intelligence 23, 123--154, 1984

....framework in which constraints on intentions, such as persistence, can be expressed. This work allows nondeterministic models in which it formalizes know how. It shows how an agent with the right knowhow can succeed with an intention if he selects actions from among those that will lead to success [Newell, 1982]. Georgeff and Rao, 1995] formalize some properties of intention maintenance in a qualitative framework. They assume full determin23 ism, and do not consider actions. However, they consider the interesting special case where an agent has only one intention in which some inferences can be drawn ....

....case where an agent has only one intention in which some inferences can be drawn qualitatively, and without considering actions. van Linder et al. 1996] formalize various attitudes including commitments. They capture a definition of commit to that recalls the action selection condition of [Newell, 1982]. Although the above approaches are useful contributions, in lacking a quantitative and explicitly normative stance, they fail to address crucial components of intention dynamics. As we argued, qualitative approaches cannot accommodate all of the three intuitive problems described in section 2. ....

Allen Newell. The knowledge level. Artificial Intelligence, 18(1):87-- 127, 1982.

....is a useful definition for agents in general. However, in many cases of interest, the agent metaphor is the most applicable when the agents are given high level cognitive specifications. This is described as taking an intentional stance toward agents [14] or viewing them at the knowledge level [17]. The high level cognitive specifications take the form of concepts such as beliefs, knowledge, desires, and intentions. They are high level, because they enable us to define the current state of an agent, what the agent might do, and how the agent might behave in different situations without ....

Allen Newell. The knowledge level. Artificial Intelligence, 18(1):87--127, 1982.

....transformations to the database, say for optimization purposes, 2 will also affect the results of revision. The resulting extreme dependence of revision on syntactic representation in our view clearly undermines the claim of [Neb89, Han96] that syntax based revision provides a knowledge level [New82] account of revision. Arguments for syntax dependence in essence boil down to the practical convenience of using the database axiomatization as a device for specifying a foundational revision policy, that is, the set of basic beliefs. 3 This convenience is unaffected by our reformulation, ....

Allen Newell. The knowledge level. Artificial Intelligence, 18:87-- 127, 1982.

....The point is that perfectly rational behaviour is a well defined function of E and U, which I will call the task environment. The problem of computing this function is addressed below. The theoretical role of perfect rationality within AI is well described by Newell s paper on the Knowledge Level [Newell, 1982] . Knowledge level analysis of AI systems relies on an assumption of perfect rationality. It can be used to establish an upper bound on the performance of any possible system, by establishing what a perfectly rational agent would do given the same knowledge. Although the knowledge that a perfectly ....

....in principle rather than in practice. 2 Calculative rationality has been the mainstay of both the logical and the decisiontheoretic traditions. In the logical tradition, the performance measure accepts behaviours that achieve the specified goal in all cases and rejects any others. Thus Newell [1982] defines rational actions as those that are guaranteed to achieve one of the agent s goals. Logical planning systems, such as theorem provers using situation calculus, satisfy the conditions of calculative rationality 2 Perhaps not coincidentally, this decision was taken before the question of ....

Allen Newell. The knowledge level. Artificial Intelligence, 18(1):82--127, 1982.

....is required for planning as well as for diagnosis tasks, in order to achieve more powerful and domain tailored case based problem solvers. From the current experience in knowledge acquisition for traditional knowledge based systems, we can at least distinguish two different types of knowledge [Newell, 1982]: Domainknowledge and problem solving knowledge [Wielinga et al. 1992] Problem solving knowledge describes the process of problem solving in terms of steps (i.e. basic inferences or subtasks as in KADS) that should to be executed to (efficiently) derive a solution. This kind of knowledge is the ....

Allen Newell. The knowledge level. Artificial Intelligence, 18:87--127, 1982.

....is best served by explicitly representing in the data structures of a program as much as possible of what a system needs to know. Allen Newell s later, and also influential, idea of a knowledge level , for which symbol systems would offer only concrete implementations, goes along the same lines [ Newell, 1982 ] Newell, 1990 ] It is worth quoting [ Levesque and Brachman, 1987 ] on the role of symbolic systems: the symbolic structures within a knowledge based system must play a causal role in the behavior of that system, as opposed to, say, comments in a programming language. Moreover, the influence ....

Allen Newell. The knowledge level. Artificial Intelligence, 18:87--127, 1982.

....has a morbid sensitivity to the syntax of the description of the world [ Winslett, 1988 ] which is considered as an undesirable property. It is argued that revision shall be independent from the syntactical representation of a belief set, that they should be specified on the knowledge level [ Newell, 1982 ] Dalal [ 1988 ] formulated the principle of irrelevance of syntax which states that a revision shall be independent of the syntactic form of the belief base representing a belief set and of the syntactic form of the sentence that has to be incorporated into the belief set (see also [ Katsuno ....

Allen Newell. The knowledge level. Artificial Intelligence, 18(1):87--127, 1982.

....corresponding trajectories are required to be defined over the same subset of F . The construction of a history R, for a given chronicle Upsilon and a given valuation M , is defined using a game between an ego (corresponding to the knowledge level of an intelligent agent proposed by Newell in [23]) and a world that comprises its physical environment and the sensori motoric system of the robot itself. This game represents the higher level construct above the level of the trajectories. It is reminiscent of the agent model of Genesereth and Nilsson [11] 4.2 The ego world game It is ....

Allen Newell. The knowledge level. Artificial Intelligence, 18(1):87--127, 1982.

....of definitions of intentions of teams that combine aspects of previous work on intentions, coordination, social commitments, and structure. Social Stance. Two powerful ways of looking at agency are the intentional stance due to Dennett [6] and McCarthy [15] and the knowledge level due to Newell [16]. These approaches legitimize the ascription of intentions to complex physical systems. We implicitly adopt these approaches in defining the intentions of teams. Traditionally, the intentional stance supervenes of the design stance, and the knowledge level on the symbol level. However, in the case ....

Allen Newell. The knowledge level. Artificial Intelligence, 18(1):87--127, 1982.

....and generates a behaviour which in turn causes the environment to generate a state history. The performance measure evaluates the state history to arrive at the value of the agent. The theoretical role of perfect rationality within AI is welldescribed by Newell s paper on the Knowledge Level [ Newell, 1982] . Knowledge level analysis of AI systems relies on an assumption of perfect rationality. It can be used to establish an upper bound on the performance of any possible system, by establishing what a perfectly rational agent would do given the same knowledge. The questionof learning in perfectly ....

....on the task environment to render decision problems tractable. This methodology has been pursued in both the logical and the decision theoretic traditions. In the logical tradition, the performance measure accepts behaviours that achieve the specified goal in all cases and rejects any others. Thus Newell [1982] defines rational actions as those that are guaranteed to achieve one of the agent s goals. Logical planning systems, such as theorem provers using situation calculus, satisfy the conditions of calculative rationality under this definition. In the decision theoretic tradition, the design of ....

Allen Newell. The knowledge level. Artificial Intelligence, 18(1):82--127, 1982.

....surprises. 2.1.1 The Structure of an IDS on the Image Level The structure of an IDS is described in fig 2.1. It consists of one or more agents and a material system M. An agent has two parts: an ego K where the reasoning capabilities of the agent resides, corresponding to Newell s knowledge level [48]; and a vehicle V which is the physical and pre symbolic interface between the ego and the reality (the material system and the other agents) The vehicle is typically equipped with perceptors (e.g. sensors, cameras) and actuators (e.g. robot arms, wheels, speakers) Relative to an ego, the world ....

Allen Newell. The knowledge level. Artificial Intelligence, 18(1):81--132, 1982.

....the assumptions implicit in certain representation shifts that are used in real learning systems, and 3. to suggest a methodology for selection of an appropriate representation shift, given some background knowledge about the learning problem. All analysis is done at the knowledge level [ Newell, 1982 ] consideration is given only to when learnability is made possible or impossible, not to when it is made easy or difficult. This enables our results to be independent of particular learning algorithms and particular learnability criteria; however, it also restricts the analysis to ....

Allen Newell. The knowledge level. Artificial Intelligence, 18:87--127, 1982.

No context found.

Allen Newell. The knowledge level. Artificial Intelligence, 18:87--127, 1982.

No context found.

Allen Newell. The knowledge level. Artificial Intelligence, 18(1):87-- 127, January 1982.

No context found.

Allen Newell. The knowledge level. Artificial Intelligence, 18(1):81--132, 1982.

No context found.

T.Dean et al., Artificial Intelligence, The Benjamin/Cummings, 1995.

No context found.

Dean, T., Allen, J., and Aloimonos, Y. (1995). Artificial Intelligence. Theory and practice. Addison-Wesley Pub., Redwood City, CA.

No context found.

Allen Newell, The Knowledge Level, Artificial Intelligence 18 (1982) 87-127.

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