Showing papers in "Artificial Intelligence in 1982"

Abstract: The Rete Match Algorithm is an efficient method for comparing a large collection of patterns to a large collection of objects. It finds all the objects that match each pattern. The algorithm was developed for use in production system interpreters, and it has been used for systems containing from a few hundred to more than a thousand patterns and objects. This article presents the algorithm in detail. It explains the basic concepts of the algorithm, it describes pattern and object representations that are appropriate for the algorithm, and it describes the operations performed by the pattern matcher.

Abstract: : This paper focuses on a basic substantive problem: the nature of knowledge and representation There are ample indications that artificial intelligence is in need of substantial work in this area The paper proposes a theory of the nature of knowledge, namely, that there is another computer system level immediately above the symbol (or program) level The nature of computer system levels is reviewed, the new level proposed, and its definition is treated in detail Knowledge itself is the processing medium at this level and the principle of rationality plays a central role Some consequences of the existence of the knowledge level and some relations to other fields are discussed

Abstract: The problem of concept learning, or forming a general description of a class of objects given a set of examples and non-examples, is viewed here as a search problem. Existing programs that generalize from examples are characterized in terms of the classes of search strategies that they employ. Several classes of search strategies are then analyzed and compared in terms of their relative capabilities and computational complexities.

Abstract: R1 is a program that configures VAX-11/780 computer systems. Given a customer's order, it determines what, if any, modifications have to be made to the order for reasons of system functionality and produces a number of diagrams showing how the various components on the order are to be associated. The program is currently being used on a regular basis by Digital Equipment Corporation's manufacturing organization. R1 is implemented as a production system. It uses Match as its principal problem solving method; it has sufficient knowledge of the configuration domain and of the peculiarities of the various configuration constraints that at each step in the configuration process, it simply recognizes what to do. Consequently, little search is required in order for it to configure a computer system.

Abstract: $&% ' () * + \" , .*/ 0 . %% . 1 /. 1 '2 34 5 6 7898 : 4 ; 9 < =4 5 6 < > ?: 8 @BA C D EE F GIH EJJ KL E G M N E H M O PRQ S T C D E H EJMU H VEJ S ETET KS M H TE H CM GIH VEJJ W V X JEJM O X Q S Y UQ Y EJ Z:; [ 7 : \\8 : ==4 5 6 < > ?: 8 @ ND QC ] Q ^ EJ KS T KLK TUQ X VM S JC H VC K M S J M H JE S CE S VEJ ] M H ET K_ VU X CCMU S TE H JCQ S T `DK J K JC D ETM Y a C D QC N M HHK ETC D VQC a C D QC ^KXX ETC D E H QC a C D QQCEC D E ] Q X C a C D QC X Q b KS C D E D JEC D QC c QV ^ dU KX C e `DK J K JC D E ] Q X CC D CC D E H CC D CC D VQCC D CC D ETM Y N M HHK ET ^KXX TQCE e

Abstract: This is a tutorial about the organization of expert problem-solving programs. We begin with a restricted class of problems that admits a very simple organization. To make this organization feasible it is required that the input data be static and reliable and that the solution space be small enough to search exhaustively. These assumptions are then relaxed, one at a time, in case study of ten more sophisticated organizational prescriptions. The first cases give techniques for dealing with unreliable data and time-varying data. Other cases show techniques for creating and reasoning with abstract solution spaces and using multiple lines of reasoning. The prescriptions are compared for their coverage and illustrated by examples from recent expert systems.

Abstract: A method for recovering the three-dimensional structure of moving rigid and jointed objects from several single camera views is presented. The method is based on the fixed axis assumption: all movement consists of translations and rotations about an axis that is fixed in direction for short periods of time. This assumption makes it possible to recover the structure of any group of two or more rigidly connected points. The structure of jointed objects is recovered by analyzing them as collections of rigid parts, and then unifying the structures proposed for the parts. The method presented here has been tested on several sets of data, including movies used to demonstrate human perception of structure from motion.

Abstract: This paper describes the design and implementation of CO-OP, a natural language database query system that provides conversationally cooperative responses to natural language requests for data retrieval and operates with a typical CODASYL database management system A conversationally cooperative response to a question is one which addresses the goals and intentions of the questioner while observing the conventions normally associated with discourse, to demonstrate that the chosen approach captures some fundamental properties of conversational cooperation, CO-OP was designed to be portable to new database domains with a minimum of effort—the lexicon and database itself are the only sources of domain specific knowledge A limited theory of cooperative behavior in question-answering is informally presented, and a practical technique for applying this theory in a natural language query system is described Sample transcripts of the system that illustrate cooperative responses in two contrasting applications are given

Abstract: Much learning is done by way of studying precedents and exercises. A teacher supplies a story, gives a problem, and expects a student both to solve a problem and to discover a principle. The student must find the correspondence between the story and the problem, apply the knowledge in the story to solve the problem, generalize to form a principle, and index the principle so that it can be retrieved when appropriate. This sort of learning pervades Management, Political Science, Economics, Law, and Medicine as well as the development of common-sense knowledge about life in general. This paper presents a theory of how it is possible to learn by precedents and exercises and describes an implemented system that exploits the theory. The theory holds that causal relations identify the regularities that can be exploited from past experience, given a satisfactory representation for situations. The representation used stresses actors and objects which are taken from English-like input and arranged into a kind of semantic network. Principles emerge in the form of production rules which are expressed in the same way situations are.

Abstract: The proof procedure we describe operates on quantifier-free formulas of the predicate calculus which are not truth-functionally normalized in any way. The procedure involves a single inference rule called NC-resolution, and is shown to be complete. Completeness is also obtained for a simple restriction on the rule's application. Examples are given using NC-resolution to derive a logic program from its specification, and to 'execute' a program specification in its original form.

Abstract: Game trees are a useful model of many kinds of decision-making situations, and have been the subject of considerable investigation by researchers in both artificial intelligence and decision analysis. Until recently it was almost universally believed that searching deeper on a game tree would in general improve the quality of a decision. However, recent theoretical investigations [8–10] by this author have demonstrated the existence of an infinite class of game trees for which searching deeper consistently degrades the quality of a decision. This paper extends the previous work in two ways. First, the existence of pathology is demonstrated in a real game (Pearl's Game) using a real evaluation function. This pathological behavior occurs despite the fact that the evaluation function function increases dramatically in accuracy toward the end of the game. Second, the similarities and differences between this game and a related nonpathological game are used as grounds for speculation on why pathology occurs in some games and not in others.

Abstract: We present a distributed algorithm for implementing α-β search on a tree of processors. Each processor is an independent computer with its own memory and is connected by communication lines to each of its nearest neighbors. Measurements of the algorithm's performance on the Arachne distributed operating system are presented. A theoretical model is developed that predicts at least order of k12 speedup with k processors.

Abstract: In this paper, ideas are presented for the expression of natural language grammars in clausal logic, following the work of Colmerauer, Kowalski, Dahl, Warren, and F. Pereira. A uniform format for syntactic structures is proposed, in which every syntactic item consists of a central predication, a cluster of modifiers, a list of features, and a determiner. The modifiers of a syntactic item are again syntactic items (of the same format), and a modifier's determiner shows its function in the semantic structure. Rules for semantic interpretation are given which include the determination of scoping of modifiers (with quantifier scoping as a special case). In the rules for syntax, the notions of slots and slot-filling play an important role, based on previous work by the author. The ideas have been tested in an English data base query system, implemented in Prolog.

Abstract: This paper proposes a general production system architecture that allows procedural control knowledge to be directly represented and used. This architecture, called a controlled production system, is based on a separately specified control structure that effects control over production invocation and interaction independently of the search strategy. It is shown that a controlled production system provides a basis for describing and implementing control constructs which, unlike most existing schemes, is formally adequate and retains all the properties desired of a knowledge based system—modularity, flexibility, extensibility and explanatory capacity. We also show that this architecture provides for a uniform programming methodology—the procedural languages and the declarative languages turn out to be special cases of a controlled production system. Schemes for improving system efficiency and resolving nondeterminism are also examined. It is shown that the separate representation of control provides a basis for a theory of efficiency transformations on production systems, and allows for more effective means of directing search.

Abstract: This paper describes two algorithms for finding the optimal interpretation of an unknown utterance in a continuous speech understanding system. These methods guarantee that the first complete interpretation found will be the best scoring interpretation possible. Moreover, unlike other optimal strategies, they do not make finite-state assumptions about the nature of the grammar for the language being recognized. One of the methods, the density method, is especially interesting because it is not an instance of the ‘optimal’ A∗ algorithm of Hart, Nilsson, and Raphael, and appears to be superior to it in the domains in which it is applicable. The other method, the shortfall method, is an instance of the A∗ algorithm using a particular heuristic function. Proofs of the guaranteed discovery of the best interpretation and some empirical comparisons of the methods are given. The relationship of these methods to strategies used in existing speech understanding systems is also discussed. Although presented in the speech context, the algorithms are applicable to a general class of optimization and heuristic search problems.

Abstract: Some new approaches to mechanical theorem proving in the first-order predicate calculus are presented. These are based on a natural deduction system which can be used to show that a set of clauses is inconsistent. This natural deduction system distinguishes positive from negative literals and treats clauses having 0, 1, and 2 or more positive literals in three separate ways. Several such systems are presented. The systems are complete and relatively simple and allow a goal to be decomposed into subgoals, and solutions to the subgoals can then be searched for in the same way. Also, the systems permit a natural use of semantic information to delete unachievable subgoals. The goal-subgoal structure of these systems should allow much of the current artificial intelligence methodology to be applied to mechanical theorem proving.

Abstract: : Learning is defined to be the computation done by a student when there is a transfer of information to him from a teacher. In the particular kind of learning discussed, the teacher names a source and a destination. In the sentence, 'Robbie is like a fox,' fox is the source and Robbie is the destination. The student, on analyzing the teacher's instruction, computes a kind of filter called a transfer frame. It stands between the source and the destination and determines what information is allowed to pass from one to the other. Computing the transfer frame requires two steps: hypothesis and evaluation. In the hypothesis step, potentially useful transfer frames are produced through an analysis of the information in the source and its immediate relatives. For Robbie, a robot, the way it compares with other robots would be noted. In the evaluation step, the better of the hypothesized frames are selected through a study of the destination frame, its relatives, and the general context. Some source-destination pairs may be generated by the student acting alone. There is also the possibility of making notes that are useful in deciding if conclusion makes sense. (Author)

Abstract: In this paper three algorithms for testing the complementarity of a matrix (representing a propositional formula) are developed in stages. Any of these algorithms is distinguished from its predecessor by a specific feature (linearity, jump, non-normal form) which endows it with a provable advantage w.r.t. its performance. For well-known proof procedures it is shown that they can be simulated by at least one of these algorithms which provides a hierarchy among several ATP-methods.

Abstract: PARADISE (PAttern Recognition Applied to DIrecting SEarch) uses a knowledge-based analysis and little searching to find the correct move in chess middle game positions. PARADISE's search does not have a depth limit or any other artificial effort limit. This paper describes the methods used to constrain the search. The ideas of using different strategies to show that one move is best and using ranges to express the values of moves (first developed in Berliner's B* search), are extended and clarified. PARADISE combines these ideas with the use of plans, a threshold, and various measures of possibility. Examples are presented, including one in which PARADISE uses an indirect strategy to prove that one move is best without finding the winning line (a first for a chess program).

Abstract: Game trees are an important model of decision-making situations, both in artificial intelligence and decision analysis, but many of the properties of game trees are not well understood. One of these properties is known as biasing: when a minimax search is done to an odd search depth, all moves tend to look good, and when it is done to an even search depth, all modes tend to look bad. One explanation sometimes proposed for biasing is that whenever a player makes a move his position is 'strengthened', and that the evaluation function used in the minimax search reflects this. However, the mathematical results in this paper suggest that biasing may instead be due to the errors made by the evaluation function.

Abstract: Cognitive psychology and artificial intelligence are in many respects indistinguishable (a claim I argued in Pylyshyn (1979)). Furthermore cognitive psychologists are extremely plentiful and are professionally a very productive breed. Consequently it requires more than a little gall and pretention to set out to comment on some papers that AI people may find useful: There is no doubt someone who would find something of interest in almost every paper in the field. The best thing I could do then is to simply list all the best papers and recommend them to you. But not even I am so foolhardy as to pretend to that task. All I can do by way of a rather unsatisfactory compromise is to give you a very abbreviated skim based solely on what I happen to have been thinking about recently, which is naturally conditioned not only by my biased view of the field but also by what research I have been interested in this week. That accounts for why some of my own papers are on this list: its not that they are particularly good, it is just that I tend to think about them more frequently than some other papers! First a general remark about sources. Cognitive psychology research is reported in very many journals inasmuch as it covers a wide range of topics, from cognitive neuroscience to cognitive anthropology, including various specialized areas such as vision, audition, psychophysics, speech, psycholinguistics, human-machine interactions, education, and so on, each of which has its own special journals. There are a number of journals, however, that are devoted to reporting research in the general area of cognition-especially as it is practiced in the approach known as 'cognitive science'. Among those that the reader should look at in searching for psychological ideas are Cognitive Psychology (the second most cited journal in all of psychology!), Cognition, Cognitive