McAllester D., "Truth Maintenance", proceedings of AAAI90.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....systems area also has contributed to improving backtracking. Stallman and Sussman [75] were the rst to mention nogood recording, and their idea gave rise to look back type algorithms, called dependency directed backtracking algorithms, that include both backjumping and nogood recording 44 [53]. In the context of solving propositional satis ability, Logemann, Davis and Loveland [16] introduced a backtracking algorithm (DLL) that uses lookahead for variable selection in the form of unit resolution, which is similar to arc consistency. To date, this algorithm is perceived as one of the ....

D. A. McAllester. Truth maintenance. In AAAI-90: Proceedings of the Eighth National Conference on Arti cial Intelligence, pages 1109-1116, 1990.

....For example, the scene labeling scheme of Waltz [28] often leads to globally consistent objects, although each step examines only neighboring edges and vertices. Truth maintenance systems [9] often sacrifice completeness by limiting the inferential steps to those involving constraint propagation [20, 21]. Such local techniques share the computational advantages mentioned earlier: there are only a few data items participating in each inference step, these items bear meaningful conceptual relationships to one another, partial results are stored exactly where they will be useful, computational steps ....

D.A. McAllester, Truth maintenance, in: Proceedings AAAI-90, Boston, MA (1990) 1109- 1115.

....so we refer the reader interested in such to [37] and confine the following to more relevant work. 7.1. Plausible inference TNT s inference mechanism adds an evidential reasoning algorithm to a forwardchaining rule engine that itself is coupled to a logic based truth maintenance system, or LTMS [16,32,33]. Several other researchers have combined truth maintenance and probabilistic reasoning systems. Falkenhainer has added the Dempster Shafer belief functions to a justification based TMS (JTMS) to develop a belief maintenance system [13] For our purposes, the JTMS, which is limited to ....

....most probable construal. Plausible reasoning, however, is also subject to exponential behavior. The obvious example is the joint probability distribution, a simple (but exponential) enumeration of the probability of every possible state of the world in a table. Bayesian belief networks (BBNs) [33] have evolved as a means for managing this intractability. They are based on the insight that domains tend to have an underlying causal structure. Carefully crafted representations of this structure via directed acyclic graphs can dramatically reduce the number of probability numbers one must ....

D.A. McAllester, Truth maintenance, in: Proceedings Eighth National Conference on Artificial Intelligence (AAAI-90), Boston, MA, Morgan Kaufmann, San Mateo, CA, 1990.

....as Dynamic Constraint Satisfaction Problem (DCSP) by Dechter and Dechter in [DD88] The concept of DCSP was introduced in context of Truth Maintenance Systems [Doy81] but can be easily adopted to Constraint Satisfaction Problem. On the relation between Truth Maintenance System and CSP see [McA90, dK89] 9 De nition 2.14 A Dynamic Constraint Satisfaction Problem is a sequence of CSPs, where every CSP is a result of changes in the preceding one. A di erence between two consecutive CSPs is expressed by the set of constraints added C add and constraints deleted from the problem. C del . ....

D. McAllester. Truth maintenance. In R. Smith and T. Mitchell, editors, Proceedings of the Eighth National Conference on Arti cial Intelligence, volume 2, pages 1109-1116. AAAI Press, 1990. 45

.... not really necessary , reached (a) as a result of fine grained interactions aiming at subsequent refinements of partial ontologies, b) where each ontology consists simply of terms and constraints in a Logic that is a proper subset of Propositional Logic, called Boolean Constraint Propagation [6]. It is this Logic that is considered acquired by the partners in the conceptualization process: as it is by far simpler, the approach has shown relevant advantages with respect to other ones. The significant advantage of the approach is that being it constructive it is founded on ....

David McAllester, `Truth maintenance', in Proc. of AAAI-90, pp. 1109-- 1116, Boston, MA, (1990).

....for the KR system. An incomplete reasoner may fail to infer some information warranted by the underlying logic, while an unsound 1 reasoner may infer information that is not warranted by the underlying logic. Typical examples of this approach include Boolean Constraint Propagation (BCP) McA80, McA90] a variant of unit resolution [CL73] and tautological entailment [Bel77, Lev84b, Fri87] both of which are sound but incomplete. The general difficulty with this approach is in characterizing (preferably syntactically) the class of queries that will be answered correctly, or the degree of error ....

.... case are known to be tractable (i.e. provably in PTIME) There are many applications where reasoning with non clausal theories is required, for example, in verifying automatically generated update constraints [GTT92] and in the applications of logical Truth Maintenance Systems (TMS) de 90, McA90, Mar91] We develop fact propagation (FP) which tractably extends BCP to non clausal theories. We present a quadratic time algorithm for FP, which runs in linear time for clausal theories. Moreover, FP is proved to be more complete than CNF BCP, a previously proposed extension of BCP to ....

[Article contains additional citation context not shown here]

D. McAllester. Truth maintenance. In Proceedings Eight National Conference on Artificial Intelligence (AAAI-90), pages 1109--1116, 1990.

....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 a theory T 1 , BCP monotonically expands T by adding literals as follows: in each step, if any single clause in T and all 1 We will assume ....

D. McAllester. Truth maintenance. In AAAI'90, pages 1109--1116, 1990.

....automated classification based on subsumption checking. The language A Box is essentially a constraint propagation engine that makes instances conform to the various constraints asserted about them. It uses a propositional representation of instances and roles, a boolean constraint propagation TMS [2] and a number of non standard services of which the following are of interest here: Rules. A rule is an implication stating that if an instance is recognized as belonging to an antecedent concept, then a consequent concept should be asserted about the instance. Rules account for an important ....

....former manager as a defined concept) as proposed in [16] which avoids the complexity of temporal subsumption verification among concepts. Even so, the existing temporal reasoning services provide a lot of leverage in reasoning. 4. 2 Temporal truth maintenance The boolean constraint propagation [2] truth maintenance system has been extended to perform label computation for time mapped propositions. Essentially, the extension allows justifications containing time intervals and temporal variables to be used. Such a justification is, for example: failure free M1) int 1) operates ....

[Article contains additional citation context not shown here]

McAllester, D. Truth Maintenance, Proceedings AAAI-90, pp. 1109-1116.

....systems area also has contributed to improving backtracking. Stallman and Sussman [SS77] were the first to mention no good recording, and their idea gave rise to look back type algorithms, called dependency directed backtracking algorithms, that include both backjumping and no good recording [McA90]. Later, Freuder [Fre82] and Dechter and Pearl [DP87, Dec90] introduced graph based methods for improving both the look ahead and the look back methods of backtracking. In particular, advice generation, a look ahead value selection method that prefers a value if it leads to more solutions as ....

D. A. McAllester. Truth maintenance. In AAAI-90: Proceedings of the Eighth National Conference on Artificial Intelligence, pages 1109--1116, 1990.

....with these tools, and more recent complexity analysis, have shown that both JTMS and ATMS functionalities are very inefficient, with ATMS exhibiting higher complexity than JTMS in both time and space. A common strategy for reducing computational complexity has been to use efficient algorithms [21] which are complete only for restricted languages (e.g. unit resolution for Horn theories) but may be incomplete in general. This paper examine another type of restriction, one based on the structure of the knowledge base as reflected in the graphical properties of the TMS constraint set. We ....

....of constraints has the same expressive power as propositional logic [20] all the algorithms presented here are applicable to propositional languages. The connection between truth maintenance systems and constraint satisfaction problems was already pointed out by several authors e.g. 6] and [23] [21]. The main thrust of these efforts has been to show that search reduction techniques developed in one area may be used to the benefit of the other. Our work here take this idea one step further. We present the algorithms in a distributed fashion, that is, assigning a processor to each variable ....

D. A. McAllester. Truth maintenance. In Proceedings of AAAI-90, pages 1109--1116, 1990.

....systems area also has contributed to improvingbacktracking. Stallman and Sussman [SS77] were the first to mention no good recording, and their idea gave rise to look back type algorithms, called dependency directed backtracking algorithms, that include both backjumping and no good recording [McA90]. Later, Freuder [Fre82] and Dechter and Pearl [DP87, Dec90] introduced graph based methods for improving both the look ahead and the look back methods of backtracking. In particular, advice generation, a look ahead value selection method that prefers a value if it leads to more solutions as ....

D. A. McAllester. Truth maintenance. In AAAI-90: Proceedings of the Eighth National Conference on Artificial Intelligence, pages 1109--1116, 1990.

....For example, the first set of clauses may capture the behavior of a device with arbitrary inputs, while the second set, typically consisting of unit clauses, represents various assumptions about the inputs of this device. For this purpose, a logical truth maintenance system (LTMS) is often used [ McAllester, 1990 ] The task of a LTMS is, given a set of clauses Sigma, to produce a labeling L Sigma for all the symbols in the language, L Sigma : P ftrue; false; unknowng, when Sigma is consistent, and to detect a contradiction otherwise. For consistent Sigma, the labeling is sound with respect to ....

McAllester, David 1990. Truth maintenance. In Proceedings of the Ninth Conference of the AAAI.

....(X 4 Y = X n Y ) Y n X) Due to the difficulty of quickly computing this form of conservative updates, however, RMS only approximated conservative updates. Reason maintenance systems developed later use simpler notions of conservatism which may be computed more rapidly (see, for example, [19, 23]. The RMS also supports the operation of the dependency directed backtracking (DDB) system, which attempts to defeat assumptions underlying those nodes identified as contradictions. Now consistency does not matter to RMS in adding or deleting reasons; instead, the basic operations of RMS only ....

D. McAllester. Truth maintenance. In Proceedings of the Eighth National Conference on Artificial Intelligence, volume 2, pages 1109--1116, Menlo Park, CA, 1990. AAAI Press.

....to reach their potential, automated tuning methods (which don t require solving complete problem instances ) must be developed; 79] reports on work in this direction. 3.3. 3 Incremental SAT Solving The problem of propositional satisfiability is closely related to that of truth maintenance [20, 77, 18]; we focus on LTMS style truth maintenance systems [76] Both problems concern a CNF formula represented as a set of clauses Sigma over a set of propositional variables V. A SAT solver seeks to find a truth assignment (i.e. a function from V to ftrue; falseg) that makes Sigma true. An LTMS has ....

D. McAllester. Truth maintenance. In Proc. 8th Nat. Conf. AI, pages 1109--1116, 1990.

.... called Logic based tmss (ltmss) These systems manipulate full propositional formulas built from propositions and truthfunctional connectives, rather than networks of dependencies, and propagate truth values using a unit resolution style algorithm called Boolean Constraint Propagation (bcp) McAllester, 1990 ] In a ltms, logical formulas are recorded in clausal form. Each clause (a finite disjunction of atomic propositions) represents a logical constraint on the truth values of its propositions. Each proposition may appear negated or unnegated in a clause; we will speak of the sign of a proposition ....

D. McAllester. Truth maintenance. In Proceedings of the 8th AAAI, pages 1109--1115, 1990.

....with some related works will be provided. 2 Ignorant Belief Networks The representation and use of incomplete information is a long standing challenge for AI researchers. During the past decade, they have developed a class of reasoning systems, called Truth Maintenance Systems (tmss) McAllester, 1990 ] which incrementally record justifications for beliefs and propagate Boolean truth values along chains of justifications. tmss are independent reasoning modules which incrementally maintain the beliefs for a general problem solver and enable it to reason on the basis of temporary assumptions ....

....of [ The lbms uses a forward chained unit resolution style algorithm called Epistemic Constraint Propagation (ecp) to propagate labels over a network of propositional formulas. ecp can be regarded as an extension to intervals of the Boolean Constraint Propagation (bcp) algorithm [ McAllester, 1990 ] used by the tmss based on the propositional calculus. The intuition behind the bcp is simple and elegant. The algorithm starts converting any formula in cnf, that is, a set of clauses. Each clause acts as a constraint on the truth values of the literals occurring in it. To be satisfied, a clause ....

D. McAllester. Truth maintenance. In Proceedings of the National Conference on Artificial Intelligence, pages 1109--1115, 1990.

....follow from a given set of constraints. In other lifting transformations, semantically complete inference requires exponential time in the worst case. Fortunately there are fast but incomplete inference techniques which seem quite effective in improving search efficiency [van Hentenryck, 1989] [McAllester, 1990]. ....

David McAllester. Truth maintenance. In Proceedings AAAI90, pages 1109-1116. Morgan Kaufmann Publishers, 1990.

....time in the size of the two formulas. The inference mechanisms employed in the VAGCAD system run in linear time but are not complete for Boolean inference. They correspond to the Boolean constraint propagation (BCP) rules that have been used by a variety of constraint processing systems [McAllester, 1990]. The BCP rules are used to derive values for expressions of type Boolean. It is important that Boolean expressions have values in the same sense that other expressions, such as numerical expressions, have values. The system has inference rules stating that an assertion Phi is in the data base ....

....Another rule states if the system has determined that (and Phi Psi) is true then it will infer that Phi is 21 true and that Psi is true. In general, BCP includes all logically valid inference rules relating a Boolean operation to the arguments of that operation. The reader is referred to [McAllester, 1990] for more details. As stated above, BCP does not derive all assertion derivable by Boolean reasoning. For example the rules will not derive Phi from (or Phi Psi) and (or Phi (not Psi) This inference, while sound, is not local it can not be derived from the local propagation of values ....

David McAllester. Truth maintenance. In Proceedings AAAI90, pages 1109--1116. Morgan Kaufmann Publishers, 1990. internet file ftp.ai.mit.edu:/pub/dam/aaai90.ps.

No context found.

David Allen McAllester. Truth maintenance. In Proceedings of the Eighth National Conference on Artifical Intelligence, pages 1109--1116. Morgan Kaufmann Publishers, Inc., 1990.

No context found.

McAllester D., "Truth Maintenance", proceedings of AAAI90.

No context found.

D. McAllester. Truth Maintenance. Proceedings of AAAI-90, pp. 1109-1116, 1990.

No context found.

D. McAllester. Truth Maintenance. Proceedings of AAAI-90, pp. 1109-1116, 1990.

No context found.

D. McAllester. Truth maintenance. In Proceedings of the Eighth (U.S.) National Conference on Artificial Intelligence, pages 1109--1116, 1990.

No context found.

D. McAllester. Truth maintenance. In AAAI-90: Proceedings 8th National Conference on Artificial Intelligence, pages 1109--1116, 1990.

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