Elgot-Drapkin, J., Perlis, D., "Reasoning situated in time I: Basic concepts", Journal of Experimental and Theoretical Artificial Intelligence 2, pp. 75-98, 1990.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....from (all the cosis in) w ff would be those having the following property: 8w 0 ffl w ff , 8w 00 ffl W , w 0 R 0 w 00 ) f P, Q R) P Q) g w 00 21 Some exceptions were already mentioned in section 2.2. Nirkhe et al. see [NKP94] show how step logics ( EMP90] Elgo88] [ElPe90]) may be used as a way to model the agent s ongoing process of reasoning; they even take into account the actual time that the agent consumes in its reasoning processes. Kraus and Subrahmanian develop in [KrSu95] a family of temporal logics in which belief update is captured by how the agent s ....

Elgot-Drapkin, J., Perlis, D., "Reasoning situated in time I: Basic concepts", Journal of Experimental and Theoretical Artificial Intelligence 2, pp. 75-98, 1990.

....modalities. We still have the problem, in our logic, that if an agent believes p, and q is logically equivalent to it, then the agent also believes q. However, the agent may believe p but not q, even if p q. Other approaches that attempt to solve the problems associated with omniscience include [84,38,71,33,39]. Works for appropriate semantics for intentions include [69,28,126] and [73] which also take the minimal structure approach. The advantage of our approach is that we deal with all the modalities in the same manner, which addresses both issues (omniscience and side effects for intentions) and ....

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

....are in the Appendix. Awareness is needed for negative introspection stating that an agent aware of but not knowing a formula at time t will know at the next time period that it did not know the formula at time t. This is useful in the solution of various problems such as the three wise men problem [16]. AR5g. Positive introspection: Positive introspection means that an agent that knows a formula at time t will know at the next time period that it knew the formula at time t. 8zxy(K(z; x; y) K(z; x ; k(z; x; y) AR5h. Negative introspection: 8zxy(A(z; x; y) K(z; x; y) K(z; x ; neg(k(z; ....

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

....for our needs, we turn, instead, to a modification of first order logic, called step logic, that deals explicitly with the passage of time as an agent reasons and allows us, formally, to introduce observations. 25 In the formalism of step logics, introduced by Elgot Drapkin, Miller, and Perlis ([64, 24, 23]) inferences are parameterized by the time taken for their inference, and these time parameters themselves can play a role in the specification of the inference rules and axioms. 26 Step logics offer a natural representation of the evolving process of reasoning itself. A step is a fundamental ....

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

.... an appropriate way to send Metroliner to NY E Act: should I do that now or after I send Bullet through O Act: is it there now 5 Implementation We have begun to implement this approach to dialogue representation within a new dialogue manager, using Active Logic [ Gurney et al. 1997, Elgot Drapkin and Perlis, 1990 ] The dialogue manager and reasoner are relatively domain and system independent, relying however on translation actions to convert between the format used by external language and domain modules and the internal logic. The new implementation of active logic (described briefly in [ Purang et ....

.... structure can be for communication with other modules or with the user (starting from the P level and working around to the D level) In addition, similar actions can be undertaken when faced with contradictions, using Active Logic s capabilities for detecting and resolving contradictions [ Elgot Drapkin and Perlis, 1990 ] 14) shows the rule used to trigger disambiguation of any objects not given a proper name in the I req when converting to D req. 13) fif(and(compute dreq(ID) ireq(ID,obj(ID,Obj) ireq(ID,lex(Obj,null) conclusion(dreq(ID,disambiguate(ID,Obj) Although the algorithm presented here is ....

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

.... to the formalization of Grice s Manner Maxims have been proposed in Marcu and Hirst s work [Marcu and Hirst, 1996] where pragmatic inferences, having to do with these maxims are treated in terms of their Stratified Logic [Marcu and Hirst, 1995] Other logic formalism, like Perlis s Active Logic [Elgot Drapkin and Perlis, 1990] were proposed to handle implicatures that assume the Quality and Relation maxims. To our knowledge, no other formalism considers all maxims and relates them to a formal theory of contexts. 5 Context Specialization and Generalization The goal of the path finding algorithm is to reveal the web ....

Elgot-Drapkin, J. and Perlis, D. (1990). Reasoning situated in time: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2:75--98.

....in time. Dealing with nonmonotonic aspects would require writing problem dependent axioms stating what changes and what doesn t from one viewpoint to the next in time. In this respect some of the ideas presented in connection with the step active logics by Perlis and co authors look promising [22, 19]. As an example we could exploit their approach to resolving contradictions, once they manifest explicitly, by failing to inherit to the next step contradicting assumptions. This can be done since consistency of viewpoints is an option: the axiom :in(false; vp) does not hold in general; a ....

J. J. Elgot-Drapkin, D. Perlis (1990). Reasoning Situated in Time I: Basic Concepts, Journal of Experimental and Theoretical Artificial Intelligence.

....in time. Dealing with nonmonotonic aspects would require writing problem dependent axioms stating what changes and what doesn t from one viewpoint to the next in time. In this respect some of the ideas presented in connection with the step active logics by Perlis and co authors look promising [18, 15]. As an example we could exploit their approach to resolving contradictions, once they manifest explicitly, by failing to inherit to the next step contradicting assumptions. This can be done since consistency of viewpoints is an option: the axiom :in(false; vp) does not hold in general; a ....

J. J. Elgot-Drapkin, D. Perlis (1990). Reasoning Situated in Time I: Basic Concepts, Journal of Experimental and Theoretical Artificial Intelligence.

....must be made to envision a state of the world that will result from successful execution of his actions. These are indeed fanciful or empty assumptions (Def. 3b) that he must revise nonmonotonically in the face of evidence to the contrary. The underlying mechanism is a formal active logic [6, 24]. In this sense, Dudley s active mind can use its formal declarative machinery to take leaps of imagination and the term formal real time imagination could aptly describe Dudley s behavior in real time using active logics. Temporal projection is an essential component of real time planning. A ....

....versions of the infamous Yale shooting problem [10, 7, 8, 12, 15, 20, 23, 25, 2] Section 5 describes Dudley s real time treatment of the qualification and ramification problems. Section 6 compares the active logic approach to other research in temporal reasoning. 2 One Step At a Time Step logics [6] were introduced to model a common sense agent s ongoing process of reasoning in a changing world. A step logic is characterized by a language, observations and inference rules. A step is defined as a fundamental unit of inference time. Beliefs are parameterized by the time taken for their ....

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

....Bel77b] has done fundamental work in dealing with inconsistent systems of logic. Some work has also been done by Baral, Kraus, Minker and Subrahmanian [BKMS91] and Grant and Subrahmanian [GS90] to handle inconsistent nonmonotonic theories. Step logic, developed by Elgot Drapkin and Perlis [EDP90], is another approach to dealing with nonmonotonic inconsistent theories. Blair and Subrahmanian [BS89] show how to reason formally in systems containing inconsistencies. Kifer and Krishnaprasad [KK89] develop a foundation for inheritance hierarchies based on these logics of inconsistency. ....

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time i: basic concepts. J. of Experimental and Theoretical Artificial Intelligence, 2:75--98, 1990.

....behavior. But once we have this, we then need to see how such a burst can come about, i.e. what the underlying behavior is. This is what has usually been thought of as an implementation issue, and what we shall argue is anything but that. The first active logics studied were step logics (see [ 22, 27 ] 5 While one goal was to make formal commonsense reasoning more realistic (e.g. respecting temporal limitations) this was by no means the only goal; certain kinds of problems do not appear to admit of static representation, let al..one static solution. In most active logics studied to date, ....

....such as consistency and completeness. The agent theory, on the other hand, may be inconsistent and incomplete; indeed if the agent is an ordinary fallible reasoner it will be so. The two theories together form an active logic pair. A notion of completeness for the meta theory was defined in [ 22, 27 ] and is repeated here: Definition 2.1 A meta theory SL n is analytically complete, if for every positive integer i, and every constant ff naming an agent wff of the corresponding agent theory, either SL n K(i; ff) or SL n :K(i; ff) 15 In [ 18 ] we showed that our SL 0 formalism is ....

[Article contains additional citation context not shown here]

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

....the contradictants and take appropriate actions including reinstating the belief judged correct (if any) In this sense, active logic formalisms mirror common sense agents in handling contradictions. We have explored a number of aspects of active logics in some detail, from the basic concepts [ Elgot Drapkin and Perlis, 1990 ] to deadline reasoning [ Kraus et al. 1990, Nirkhe, 1994, Nirkhe et al. 1997 ] to reasoning about others ongoing reasoning ( Elgot Drapkin, 1988, Elgot Drapkin, 1991 ] and have begun to explore uses of active logics in dialogue, especially in pragmatic inferences that support dialog ....

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

....and plans for embedding that within a dialogue system. 1 Introduction In this paper, we present an approach toward design of a rational agent, integrating aspects of theoretical reasoning, practical reasoning, and reasoning about and executing plans. The approach, based on Active Logic [ Elgot Drapkin and Perlis, 1990 ] couples a view of belief as resulting explicitly from inference (or observation) with a resource bounded approach to inference. Thus not all consequences of an agent s beliefs will be believed (currently) and doing the inference necessary to establish these consequences as beliefs will take ....

....that specifies an environment in which the logic runs . Thus an active logic is not pure formalism but is a hybrid of formal system and embedded inference engine, where the formal behavior is tied to the environment via the observations and the internal monitoring of timepassage (see [ Elgot Drapkin and Perlis, 1990 ] for a detailed description) In the above example, the reactivity is not to external events but rather to the universal event of time passage vis a vis one s own reasoning. One can conceptualize this externally in terms of looking at a clock but this is not necessary or particularly helpful. On ....

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

No context found.

Elgot-Drapkin, J. and Perlis, D. 1990. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98.

No context found.

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

....and also certain covert behaviors such as hiding or watching. In these of course time is important. But there is another angle that is not usually noted, one that we have been exploring for the past several years ( Drapkin and Perlis, 1986b] Drapkin and Perlis, 1986a] ElgotDrapkin, 1988] [Elgot Drapkin and Perlis, 1990] ) Namely, action also occurs in the form of mere thinking or reasoning. Moreover, the very same temporal considerations apply to this reasoning behavior. This leads us to view reasoning itself as a kind of action, with the obvious yet non trivial consequence that our reasoning goes on as the ....

....the task at hand. In [Drapkin and Perlis, 1986b] our interest in this theme is enunciated. Some preliminary results are reported in [Drapkin and Perlis, 1986a] More recently, Elgot Drapkin, 1988] gives a wider variety of technical and implementational results, many of which are published in [Elgot Drapkin and Perlis, 1990] . The aim of the current paper is to step back and re assess the conceptual basis for this work to see how it fits into broader themes and research directions. While the format of our presentation may seem unusual, we feel that it suits the purpose of this paper well. In particular, we will ....

[Article contains additional citation context not shown here]

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

....L is a triple, L; OBS; INF , where L is a firstorder language, OBS is an observation function, and INF is an inference function. We use the notation, SL n (OBS; INF ) for such a theory (the language L is implicit in the definitions of OBS and INF ) For more background on step logic, see [ED88, EDP90]. 3 The Problem We present a variation of the classic wise menproblem which was first introduced to the AI literature by McCarthy in [McC78] This version best illustrates the type of reasoning that is so characteristic of commonsense reasoners. A king wishes to know whether his three advisors ....

....long it takes for wise man #2 to reason is missing from other approaches to this problem. A similar but more complicated solution is given for the Three wise men problem in [ED88, ED91] Step logic has been used to solve other commonsense reasoning problems, including Moore s Brother Problem (see [Moo83, ED88, EDP90]) The solutions to both the Wise men problem and the Brother Problem have been implemented on an IBM PC AT using Arity Prolog (see [ED88] for details) 7 Acknowledgments We would like to thank Don Perlis, Kevin Gary, and Laurie Ihrig for helpful comments. ....

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

....how the reasoning progresses vis a vis the passage of time. Although work in temporal logic involves reasoning about time (e.g. All84, McD82, ER87] time is not treated as a crucial resource that must be carefully rationed by the agent, as it is spent in every step of reasoning. Step logics [EDP90, PEDM] were introduced as a formal apparatus to model an agent s ongoing process of reasoning. They have since been extended and renamed as active logics. In [KNP90, NKP91, NKP93] the steplogic framework is used to create an active logic based fully deadline coupled planning and reasoning mechanism ....

....Halpern discuss the possibility of capturing bounded and situated reasoning by letting the awareness set vary over time. However, no attempt has been made to systematically study and model situations where the passage of time is a critical issue. 3 The step logic approach to reasoning Step logics [EDP90] were introduced to model a common sense agent s ongoing process of reasoning in a changing world. A step logic is characterized by a language, observations and inference rules. A step is defined as a fundamental unit of inference time. Beliefs are parameterized by the time taken for their ....

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

.... of Unknown, though several possibilities come to mind, including fix point consistency checking in the style of Reiter s Default Logic (DL) and McDermott and Doyle s NML (both undecidable) circumscription (semi decidable) and the negative introspection facility of step logics (decidable) 6] [7]. 6 3.1.2 Some pleasant features One nice feature of the iisi constant approach to default reasoning is that, like circumscription, it requires no special logical connective (e.g. Delgrande s ) connective [4] 5] which is semantically distinguished from first order material implication in ....

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

....with contradictions How can a reasoning apparatus (person, robot, program) deal with contradictions There have been various proposals. Some, such as the paraconsistent logics surveyed in [1] aim to extract a trustworthy core of inferences while avoiding the contradictions. Others, such as [5, 15, 26], aim to detect and resolve contradictions. The latter are closer in spirit to the needs we are addressing here. Unlike the traditional view that abhors a contradiction and seeks at all costs to avoid such 5 and fears that CSR will come to naught (or to disaster) in their presence, the new ....

....engine are done by the engine, not by a separate module while the engine is turned off or idling. Thus self adjusting logics of confusion seem to be the order of the day. What form such logics may eventually take is far from clear. I note that OSCAR [21, 24, 22, 23] as well as active (step) logics [4, 3, 5, 16, 9] are beginnings. It is clear that human commonsense reasoning involves many conflict driven changes of belief, and that this is in need of being better understood for both cognitive and robotic purposes. ....

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

....that provides infinitely many conclusions. Inference must be fast and narrow. So some sort of guides must contain the logic, perhaps a kind of short term memory. Active logics appear to have features well suited to all these desiderata. Already they have been applied to default reasoning [1, 3, 2] , indexicals [4, 5] observations [1, 3] plans and deadlines [6, 7] contradiction detection and recovery [1, 4] and semantic shift [4, 5] While we have not applied active logics to position indexicals, we have done so for other indexicals and for meaning change in other settings, and we ....

....Inference must be fast and narrow. So some sort of guides must contain the logic, perhaps a kind of short term memory. Active logics appear to have features well suited to all these desiderata. Already they have been applied to default reasoning [1, 3, 2] indexicals [4, 5] observations [1, 3] , plans and deadlines [6, 7] contradiction detection and recovery [1, 4] and semantic shift [4, 5] While we have not applied active logics to position indexicals, we have done so for other indexicals and for meaning change in other settings, and we suspect the same techniques can be applied ....

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

....a convenient tool for executing the required inferences (as called for by our revised version of Heim s update function) as the discourse evolves through time. Keywords: presupposition, discourse, context, accommodation, active logic 1 Background on active logic and presupposition. Active logic [ Elgot Drapkin and Perlis, 1990, Perlis and Miller, 1993 ] is a family of formalisms developed for the purpose of modeling the reasoning process in a way that respects the passage of time as reasoning proceeds. That is, the reasoner may have beliefs (draw inferences) concerning what time it is (now) and these are updated ....

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

....that it be able to change usage, employ new words, change meanings of old words, and so on. The usual fixed language with a fixed semantics that is the stockin trade of AI seems inappropriate to this task. Here we do not offer a new logic per se; rather we borrow an existing one (step logic [ Elgot Drapkin and Perlis, 1990 ] Elgot Drapkin, 1991 ] and apply it to the specific issue of language change. Referenceability, to stick with Rieger s terminology, demands that the agent and therefore the agent s language have expressions available to denote expressions themselves (e.g. via quotation) and also to ....

....shown in the previous step are inherited to the current step. Beliefs at step 1 are those held before the agent s conversation with Sally and those at step 9 reflect 11 e.g. suspending the use of potentially problematic beliefs, in particular the contradictands and their consequences. See [ Elgot Drapkin and Perlis, 1990] for details. Step 1: M(j) Names( j; j; Gamma1) AM Step 2: BL(j) M(j) Sally: his leg is broken and his wife. Step 3: AM;Names( j; j; Gamma1) Contra( M(j) M(j) Agent: Impossible He isn t married. Step 4: MISID(j; fM(j) B(j)g; 2) Sally: You misidentified ....

[Article contains additional citation context not shown here]

Elgot-Drapkin, J. and Perlis, D. 1990. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75-- 98.

....keep track of those objects in an internal language 4. classify those objects into appropriate categories 5. reason about those categories, e.g. count or estimate sizes, form and apply defaults, etc. 6. revise categories and category membership as needed 3 Discussion Some of our previous work [3, 2, 6, 10, 8, 14, 11] treated rather disjointed aspects of this longrange problem. In this position paper we simply single out four key aspects of the more integrative long range task: calibration, counting, grounding, and grouping. Some aspects of the larger task are not ones we intend to address, especially those ....

J. Elgot-Drapkin and D. Perlis. Reasoning situated in time I: Basic concepts. Journal of Experimental and Theoretical Artificial Intelligence, 2(1):75--98, 1990.

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