M. Dalal. Semantics of an anytime family of reasoners. European Conference on Artificial Intelligence (ECAI'96), pages 360--364, Budapest, 1996.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....the answering agent, whereas abductive reasoning is a way in which an agent performs reasoning to serve its own goals. Some of the work described later in this dissertation on abduction and question answering is presented in Burhans and Shapiro [2001] 2. 9 Anytime Reasoning Anytime reasoning [Dalal, 1996] refers to a reasoning process whereby the quality of a solution is related to the amount of time taken to compute the solution. At any time during reasoning, an anytime reasoner can give you the best information that has been discovered up to that point. Thus, in an anytime reasoner, each ....

Mukesh Dalal. Semantics of an anytime family of reasoners. In Proceedings of the European Conference on Artificial Intelligence, pages 360--364, Budapest, 1996.

....much of the postconditions can still be guaranteed. As a result, we depart from the traditional binary yes no pre postconditions, and aim for a more gradual treatment. This paper is quite distinct from work on approximate computational methods such as Horn compilation [10] approximate deduction [9, 4], etc. These are all techniques for specific approximate computational methods. In this paper, we do not study any such single technique, but instead we aim for a general framework in which applications of these techniques can be characterised. The remainder of this paper is structured as ....

....applica tions. We believe that the methods for approximate diagnostic reasoning described in [11] is a source of such more realistic applications of gradual methods which can be captured in our framework. We should investigate how different uniform methods of approximate reasoning (e.g. [4, 9, 10]) can be modelled in our framework. Such approximate methods generally use some parameter to characterise their degree of precision (the maximum length of a clause, or a subset of the alphabet to be used, etc) We expect that such parameters are special cases of the general metric on pre and ....

M. Dalal, 'Semantics of anytime family of reasoners', in ECAI'96, pp. 360-364.

....with a very simple language and iterating over logics of increasing strength supplementing previously derived facts. Approximate Deduction Instead of modifying the logical language, approximations can also be achieved by weakening the notion of logical consequence [Schaerf and Cadoli, 1995] [Dalal, 1996]. This can be done by restricting the size of definitions that are considered or by allowing failures on parts of the vocabulary used. On the semantic Web, using the latter techniques will often even be necessary if agents that only share parts of their vocabulary. In this case approximate ....

Dalal, M. (1996). Semantics of an anytime family of reasoners. In Proceedings if the 12th European Conference on Artificial Intelligence, pages 360--364.

....and therefore we cannot simply ig nore the uninteresting parts in the computation (unless they have a special form) The propositional approximation we propose combines in a uniform way a number of these ideas. In particular, we extend the various propositional semantics proposed by Dalal [38], Levesque [105] Schaerf : Cadoli [155] and modify the logic for awareness of Fagin : Halpern [55] We start with an approximate partition (for short ap) of the set of propo sitional atoms 7rop into three sets: unknown propositions, which are denoted by unk(7rop) interesting propositions ....

....We only prefix the word approximate to each notion. For instance we say that 27 is an approximate model for A if 27 II ; A and it is a approximate counter model for A if Z ] f.A. The only thing we revise is the definition of logical consequence (Def. 2. 2) We adopt the definition used by Dalal [38] because it is more in the spirit of approximate forcing: Definition 6.1 Given an approximate partition, a formula A is an approxi mate logical consequence of assumptions U (in symbols U = A) iff every model for U is not a counter model for A. Remark. This is not a multi valued logic although ....

[Article contains additional citation context not shown here]

M. Dalal. Semantics of an anytime family of reasoners. In Proc. of the 12th European Conf. on Artificial Intelligence (ECAI-96), pages 360-364. John Wiley & Sons, 1996.

....the accuracy of the solution improves with the time used in computing the solution. Hence, anytime deduction offers a compromise between the time complexity needed to compute answers by means of approximate entailment relations and the quality of these answers. Following this idea, Dalal in [5] presents a general technique for approximating deduction problems. The starting point of its framework relies on the entailment relation bcp defined using boolean constraint propagation. Based on this relation, the author defines a family of entailment relations bcp k , which extend ....

....a unifying, logic oriented framework which handles these two approaches and that enables us to specify anytime reasoners. We have stressed on a sound and complete multi modal logic, named ARL, which generalizes and expands in several directions previous methods concerning approximate deduction [2, 5, 17] and anytime compilation [7, 15, 18] Based on this logic, we have illustrated that the framework integrates several major features: resource bounded reasoning, improvability, dual reasoning and off line processing. We believe that the results reported here are interesting and worth of further ....

M. Dalal. Semantics of an anytime family of reasoners. In W. Wahlster, editor, Proceedings of the 12th European Conference on Artificial Intelligence (ECAI-96), pages 360--364. Wiley & Sons, 1996.

....approach was already taken in [tenTeije vanHarmelen, 1996] where we used the approximate entailment relations defined in [Schaerf Cadoli, 1995] to characterise approximate diagnostic reasoning. In this paper we consider the sound but incomplete approximation of the logical entailment from [Dalal, 1996] in the context of diagnostic reasoning. The approximation of Dalal is called BCP k and is based on boolean constraint propagation. The parameter k can be increased to improve the quality (and of course also the cost) of this approximate entailment relation as will be explained in section 3. The ....

....typical behaviour and structure models are in fact functionfree and range over finite alphabets. Such first order theories can be rewritten as equivalent propositional theories using standard methods. 3 DEFINING # BCP k This section repeats some of the definitions and results from [Dalal, 1996] and [Dalal Yang, 1997] They define the family of approximate deduction relations BCP k . Boolean Constraint Propagation [McAllester, 1990] is a limited form of deduction which is a variant of unit resolution [Chang Lee, 1973] It performs limited deduction in linear time, as follows: given ....

[Article contains additional citation context not shown here]

M. Dalal. Semantics of anytime family of reasoners. In ECAI'96, pages 360--364, 1996.

....much of the postconditions can still be guaranteed. As a result, we depart from the traditional binary yes no pre postconditions, and aim for a more gradual treatment. This paper is quite distinct from work on approximate computational methods such as Horn compilation [10] approximate deduction [9, 4], etc. These are all techniques for specific approximate computational methods. In this paper, we do not study any such single technique, but instead we aim for a general framework in which applications of these techniques can be characterised. The remainder of this paper is structured as follows: ....

....realistic applications. We believe that the methods for approximate diagnostic reasoning described in [11] is a source of such more realistic applications of gradual methods which can be captured in our framework. We should investigate how different uniform methods of approximate reasoning (e.g. [4, 9, 10]) can be modelled in our framework. Such approximate methods generally use some parameter to characterise their degree of precision (the maximum length of a clause, or a subset of the alphabet to be used, etc) We expect that such parameters are special cases of the general metric on pre and ....

M. Dalal, `Semantics of anytime family of reasoners', in ECAI'96, pp. 360--364.

.... an exact proof of a classical theorem is too hard, why not looking for an approximate one To fill this gap, a number of approximation methods for logical consequence have been developed in AI: knowledge compilation [22] semantical approximations [4,15,16,21] and syntactic anytime reasoners [7,6,8]. Knowledge compilation or vivification [7,22] is a preprocessing step to transform a theory into another one for which deduction is faster: Horn theories in [22] and boolean constraints propagation in [7] So it doesn t exactly match the intuition of approximation as an incremental process. ....

.... hard, why not looking for an approximate one To fill this gap, a number of approximation methods for logical consequence have been developed in AI: knowledge compilation [22] semantical approximations [4,15,16,21] and syntactic anytime reasoners [7,6,8] Knowledge compilation or vivification [7,22] is a preprocessing step to transform a theory into another one for which deduction is faster: Horn theories in [22] and boolean constraints propagation in [7] So it doesn t exactly match the intuition of approximation as an incremental process. Syntactic approximation [7,6,8] is incremental but ....

[Article contains additional citation context not shown here]

M. Dalal. Semantics of an anytime family of reasoners. In Proc. of the 12th Eur. Conf. on Artificial Intelligence (ECAI-96), pages 360--364. John Wiley & Sons, 1996.

....ABD S 3 to ABD S 0 3 ; or more generally a shift from an ABD component Old to an ABD component New for which holds New Old. The idea of such general strategies is that if new definitions for entailment relations are available for the ABD or CBD part of our diagnosis definition (e.g. (Dalal 1996)) we only need to verify the properties of , and ae and use the above general strategies. The strategies in this section are general in the sense that they do not exploit specific properties of the behaviour model. Using such properties would be more attractive because this enables us to ....

Dalal, M. 1996. Semantics of anytime family of reasoners.

.... sound, and tractable reasoners [4, 3, 6] Our technique for generating these anytime families is based on restricting the length of the clauses used in chaining (Modus Ponens) We will also provide model theoretic semantics for the reasoners in the anytime family based on BCP [7]. We will present results of some preliminary experiments in using these reasoners for determining consistency of temporal networks based on Allen s interval algebra [1, 10] We will also present a compilation approach that approximates formulae of an expressive knowledge base by automatically ....

M. Dalal, `Semantics of an anytime family of reasoners', in Proceedings Twelveth EuropeanConferenceon Artificial Intelligence (ECAI 96), ed., W. Wahlster, pp. 360--364, Budapest, Hungary, (1996). John Wiley and Sons, Ltd.

.... Dean 1988) They are complete reasoners that provide partial answers even if stopped prematurely; the degree of completeness of the answer improves with the time used in computing the answer. They are often used for providing a quick first cut to a problem, which can be later improved. In (Dalal 1996b; 1996a) we presented a family 0 ; 1 ; of reasoners such that each k is sound and tractable, each k 1 is at least as complete as k , and each theory has a complete reasoner k for reasoning with it. Such a family is called an anytime family of reasoners, since given any reasoning task, one can presumably start ....

Dalal, M. 1996c. Semantics of an anytime family of reasoners.

No context found.

M. Dalal. Semantics of an anytime family of reasoners. European Conference on Artificial Intelligence (ECAI'96), pages 360--364, Budapest, 1996.

No context found.

Dalal, M. 1996b. Semantics of an anytime family of reasoners. In Proceedings of the European Conference on Artificial Intelligence (ECAI-96), 360--364.

No context found.

Dalal, M. 1996b. Semantics of an anytime family of reasoners.

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