J. Crawford and B. Kuipers. Algernon - A Tractable System for Knowledge-Representation. In Proceedings of the AAAI Spring Symposium on Implemented Knowledge Representation and Reasoning Systems, Palo Alto, CA.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....to be extended to permit a full search for path specialization . Such search mechanisms are both powerful and easy to use for knowledge retrieval, and can be tractable They were also used in Algernon , an inference system based on a tractable reasoning system called Access Limited Logic [2]. All the queries or assertions that can be made via the WebKB 2 interface can also be made by any application over the Web via a GET or POST HTTP request and with the same language of commands. http: www.webkb.org ontorama If the query graph and each of the statements has a tree structure ....

J. M. Crawford and B. J. Kuipers, "Algernon -- a tractable system for knowledge representation", SIGART Bulletin 2(3), June 1991, pp. 35--44.

....of one generalization should not depend on the implementation of other generalizations. This can be a somewhat di#cult task to achieve, and may require some thought on representational style. In particular, it may require more use of find or create constructs (e.g. the :forc command in Algernon [4], or the the command in KM [5] so that a path role chain s failure to find an object (because a generalization is not included) triggers a create operation to create that object. This requirement for elaboration tolerance has also been identified in software reuse, e.g. the notion of ....

J. M. Crawford and B. J. Kuipers. Algernon -- a tractable system for knowledgerepresentation. SIGART Bulletin, 2(3):35--44, June 1991.

....4 A Personal Reflection on Algernon Finally, I present here some personal reflections on Algernon. As a preface, I ve been struck by the strong similarity of the underlying approach to inference used in the four KR systems I m familiar with (KM KQL [4] CycL [8] KRL [2] and Algernon [6]) In various syntaxes, these systems all operate using the general mechanism of generating Skolem constants to denote individuals, and then applying rules to those individuals (possibly creating new individuals in the process) to infer answers to questions. I now present some more detailed ....

J. M. Crawford and B. J. Kuipers. Algernon -- a tractable system for knowledgerepresentation. SIGART Bulletin, 2(3):35--44, June 1991.

....and the solution to the n 1 challenge. 1 This project had begun in the Fall of 1992 but I worked on it to the exclusion of all else only during the time mentioned. Because of this early association with Kuipers, the early versions of IPR used his Algernon system to store and access the knowledge [47]. 147 Fetching. Any prover attempting to solve this problem will have to use a large knowledge base. This is true even if the challenge is restated so that only statements from relevant sections are in the knowledge base. The IPR framework is a step toward solving the fetching problem. IPR is ....

J. M. Crawford and B. J. Kuipers. Algernon -- a tractable system for knowledge representation. SIGART Bulletin, 2(3):35--44, June 1991.

....The KM command the behaves like the, except rather than generate an error if a matching instance cannot be found, it will create one. the thus combines the functionality of (the . and (a . This function is called find or create, also used in other KR languages such as Algernon [11]. In fact, a and the have similar semantics (assert the existence of an instance and return it) except that a does not assume any coreference between that instance and others in the KB, while the will, if such coreference is possible. If such coreference is ambiguous, e.g. two instances match a ....

.... write class membership statements in the form isa(instance, class) rather than class(instance) the latter is common in description logics) The importance of reifying classes has also been recognized in other advanced knowledge representation languages, such as CycL [14] SNePS [15] and Algernon [11]. Suppose we wanted to represent the sentence People like big hot dogs. instead. One approach would be to create the class Big Hot Dog, but in general it is cumbersome to assign a name to every such class, particularly if it is never used anywhere else. Worse, the intended class 23 The KM code ....

[Article contains additional citation context not shown here]

J. M. Crawford and B. J. Kuipers. Algernon -- a tractable system for knowledgerepresentation. SIGART Bulletin, 2(3):35--44, June 1991.

....the motion of the disk as it rolls from the top of the track to the bottom. Figure 1. 1: A problem from Resnick and Halliday s Physics textbook, Fundamentals of Physics [76] the implementation makes use of the QPC [9, 17] and QSIM [42, 43] qualitative reasoning systems, and the Algernon [10] knowledge representation system. This implementation architecture will allow us to inherit additional capabilities as they are developed for the underlying systems, such as methods for incorporating incomplete quantitative knowledge into the reasoning process [44, 38] 1.3.4 An Illustrative ....

.... Figure Understander [74, 70, 69, 73] Once the input has been processed, if the problem involves dynamic change, we use a QPC based implementation of our spatial model to solve the problem through qualitative simulation [71] If the problem involves a static scene, we can instead use an Algernon [10] rule based implementation of our spatial model to solve the problem [70, 73] The use of diagrams to describe the spatial scene serves two purposes. One is convenience. We extract a lot of spatial information from a diagram, and as the popular saying goes, a picture can be worth a thousand ....

[Article contains additional citation context not shown here]

J. Crawford and B. Kuipers. Algernon - a tractable system for knowledgerepresentation. In AAAI Spring Symposium on Implemented Knowledge Representation and Reasoning Systems, Palo Alto, CA, 1991.

....Algernon Abstract Machine 1 , a system for executing access limited logic programs, and rule based programs in general. It is based on both the WAM and the SECD machine for LISP [Kogge, 1991] We have implemented the AAM and used it as the underlying architecture of a new version of the Algernon [Crawford and Kuipers, 1991] reasoning system. One of the few drawbacks of the WAM has been the dearth of readable, understandable explanations of the WAM. It is the author s hope that this document is both readable and useful. The next section describes access limited logic and the reasoning mechanisms it uses. It is the ....

....Prolog [Warren, 1977; At Kaci, 1991] Since Prolog reasoning is resolution based and ALL reasoning involves forward chaining, backward chaining and rule continuations, the AAM differs significantly from the WAM. A LISP based implementation of the AAM has been implemented as version 3 of Algernon [Crawford and Kuipers, 1991]. This chapter is intended to serve as a reference manual for future ALL implementations. Abstract machines are difficult to describe clearly; witness the profusion of nearly incomprehensible descriptions of the WAM. In hopes of adding clarity to this document, the AAM instructions are introduced ....

Crawford, J. M. and B. J. Kuipers (1991). Algernon -- a tractable system for knowledge representation. SIGART Bulletin 2(3): 35-44, June 1991.

....is a knowledge base fragment that contains an integrated description of the information in each form of input. In the current implementation, the Figure Understander operates on the Postscript files produced by the Interviews drawing editor, and the Algernon knowledge representation language [1] is used to maintain the knowledge base. The techniques described in this paper could be easily adopted to other drawing and CAD CAM tools. 2.1 Figure Understander Architecture The Figure Understander contains a diagram processing component and a text processing component. Diagram processing ....

J. Crawford and B. Kuipers. Algernon - a tractable system for knowledge-representation. In AAAI Spring Symposium on Implemented Knowledge Representation and Reasoning Systems, Palo Alto, CA, 1991.

....already hold the object. every Putting has (object ( a Thing) destination ( a Container) is possible ( not ( the contents of (the destination of Self) 1] includes (the object of Self) 9 (add list ( triple (the destination of Self) contents (the object of Self) [2] A Getting is possible only if the source container already ; holds the desired object. every Getting has (object ( a Thing) source ( a Container) is possible ( the contents of (the source of Self) 2] includes (the object of Self) del list ( triple (the source of Self) ....

....( triple (the destination of Self) contents (the object of Self) 2] A Getting is possible only if the source container already ; holds the desired object. every Getting has (object ( a Thing) source ( a Container) is possible ( the contents of (the source of Self) [2] includes (the object of Self) del list ( triple (the source of Self) contents (the object of Self) Thus the set of doable actions are those whose is possible slot evaluates to non nil. Below, we manually enumerate all the potential actions in a situation, then ask KM to identify ....

[Article contains additional citation context not shown here]

J. M. Crawford and B. J. Kuipers. Algernon -- a tractable system for knowledge-representation. SIGART Bulletin, 2(3):35--44, June 1991.

....next statement. This challenge was in 1 This project had begun in the Fall of 1992 but I worked on it to the exclusion of all else only during the time mentioned. Because of this early association with Kuipers, the early versions of IPR used his Algernon system to store and access the knowledge [47]. 156 place at least as early as Hao Wang s work in the late 1950s [116] This has been called the n 1 problem or the theorem n 1 challenge. John Kelley s text, General Topology [77] seems to be an excellent target for this test since it is very thorough, includes the axioms of set theory ....

J. M. Crawford and B. J. Kuipers. Algernon -- a tractable system for knowledge representation. SIGART Bulletin, 2(3):35--44, June 1991.

....such as the InterViews drawing editor, and one or more sentences of descriptive text. Its output is a fragment of a knowledge base which contains an integrated description of the information in each form of input. In the current implementation, the Algernon knowledge representation language [Crawford and Kuipers 91] is used to maintain the knowledge base. The Figure Understander makes use of a spatial reasoning module, a domain model, and a picture description model, as described further below, to process the text and diagram input. The previous work in the area of integrating text and pictoral information ....

Crawford, J. and B. Kuipers. Algernon - A Tractable System for KnowledgeRepresentation. In working notes: AAAI Spring Symposium on Implemented Knowledge Representation and Reasoning Systems, 1991.

.... The third block specifies the knowledge base assertions to be made for each individual object type ( individual object semantics) and for groups as a whole ( object group semantics) All contextual information is expressed in terms of predicates in the Algernon knowledge representation language (Crawford et al. 1991). For example, under :object group semantics, we have specified that all diagram objects of the group conducting disks are members of the knowledge base class conducting disks , and that they have the property of being solid objects. Under :individual object semantics, we specify that diagram ....

.... to the Figure Understander includes a diagram produced through an object oriented drawing editor (currently, the Postscript file for a diagram produced using the Interviews drawing editor) a Picture Semantics description for the diagram, one or more sentences of descriptive text, and an Algernon (Crawford et al. 1991) knowledge base. It outputs an integrated description of the information in each form of input, diagram and text, into the knowledge base. 3.1. Diagram processing The strength of the Figure Understander lies in its diagram interpretation capabilities. As the diagram is processed, a knowledge base ....

[Article contains additional citation context not shown here]

Crawford, J. and Kuipers, B. (1991). Algernon - A Tractable System for KnowledgeRepresentation.

....at the expense of completeness. The challenge is to adopt restrictions that, in practice, have minimal effect on question answering ability. The inference control method we present for CGs is based on access limitation , which is used elsewhere in the knowledge representation community [9]. Access limitation is based on the use of access paths, which state how a value can be computed via a chain of inferences. In the context of Conceptual Graphs, an access path describes how the value of a CG node can be determined by traversing a path in the graph. Incorporating access paths into ....

....P i linking some individual X0 to other individuals xn , such that the second argument of P i is the same as the first argument of P i 1 : P 1 (X0; x 1 ) P 2 (x 1 ; x 2 ) Pn (x n Gamma1 ; xn ) where the x i s are free variables. This is a slightly more restricted definition than in [9], but one suitable for conceptual graphs. An access path denotes the set S of values for xn (the last variable in the path) for which there exists at least one value for all the other variables in the path, i.e. 8xn ( xn 2 S 9x 1 ; xn Gamma1 P 1 (X0; x 1 ) Pn (x n Gamma1 ; xn ) ....

[Article contains additional citation context not shown here]

J. M. Crawford and B. J. Kuipers. Algernon -- a tractable system for knowledgerepresentation. SIGART Bulletin, 2(3):35--44, June 1991.

....1: Lake Travis and Mansfield Dam, Austin, TX. of new landmarks by the simulator. As a consequence SQPC cannot guarantee termination. 3. 3 Implementation status SQPC is fully implemented in Lucid Common Lisp as an extension to QPC, which in turn uses the Algernon knowledge representation system [Crawford, 1991] and QSIM. We are currently experimenting with SQPC in the water flow control domain; SQPC has been run on over a dozen examples comparable to the ones shown in this paper. The runtime for these examples range from one minute to half an hour on Sun Sparc4 75. The bulk of this time is spent ....

J. Crawford. Algernon --- a tractable system for knowledge representation. SIGART Bulletin, 2(3), 1991.

No context found.

J. Crawford, B. Kuipers, Algernon: A tractable system for knowledge representation, SIGART Bull. 2 (3) (1991) 35--44.

No context found.

J. Crawford and B. Kuipers. Algernon: a tractable system for knowledge representation. SIGART Bulletin, 2(3):35--44, 1991.

....of pmodel where teq#ds # is the case. If the identity of ds can be established, then s is asserted in pmodel.This declares ds to be known in pmodel and creates the places and paths that explain s according to the axioms of the topological theory TT#E#. the rule based system Algernon [Crawford and Kuipers, 1991] as our theorem prover. In Section 8.2 (Page 40) we present an illustrative trace of the algorithm. 8.1 Partial Models In addition to a list of schemas to explain, a partial model has associated a set of objects (i.e. distinctive states, schemas, places, paths) that are known in the model. The ....

J. Crawford and B. Kuipers. Algernon: a tractable system for knowledge representation. SIGART Bulleting, 2(3):35--44, 1991.

....all the predicates in the topological theory have a last extra argu ment for a pmdodel. For instance, instead of writing at(ds, p) we write at(ds, p, praodel) 2 prooriel SThe notation x = y is an abreviation for (x pmodel) A (y pmodel) A x = y. 19Our theorem prover in this case is Algernon [Crawford and Kuipers, 1991]. 2pmodel inherits from pmodel all known objects and facts. See page 157. 2When the pmodel is understood we drop it from the predicate arguments. 155 at(ds,p, praodel) is the case when at(ds,p) is true in the partial model praodel (i.e. praodel at(ds, p) Our logic for partial models takes ....

....creates a possible extension for praodel where ds teq dsi is the case. If some of these ex tensions are consistent, then create possible extensions also creates an extension where ds is known and different from the dstates in candidates. In this last case the function returns 24In Algemon [Crawford and Kuipers, 1991] some of these rules are implemented as forward chaining (if added) rules (e.g. rules 9.18 and 9.19) and others as backward chaining (if needed) rules (e.g. rules 9.23 and 9.24) The problem with if added rules is that they could derive a large number of useless truths. For instance, if rule 9.23 ....

J. Crawford and B. Kuipers. Algernon: a tractable system for knowledge representation. SIGARTBulleting, 2(3):35 dd, 1991.

....do so, all the predicates in the topological theory have a last extra argument for a pmdodel. For instance, instead of writing at#ds; p# we write at#ds; p; pmodel#. The notation # # # is an abreviation for ## # ####### # ## # ####### # # # #. Our theorem prover in this case is Algernon [Crawford and Kuipers, 1991] . inherits from ###### all known objects and facts. See page 157. When the pmodel is understood we drop it from the predicate arguments. 155 at#ds; p; pmodel# is the case when at#ds; p# is true in the partial model pmodel (i.e. pmodel ## at#ds; p#) Our logic for partial models takes ....

....creates a possible extension for pmodel where ds # is the case. If some of these extensions are consistent, then create possible extensions also creates an extension where ds is known and different from the dstates in candidates. In this last case the function returns In Algernon [Crawford and Kuipers, 1991] some of these rules are implemented as forward chaining (ifadded) rules (e.g. rules 9.18 and 9.19) and others as backward chaining (if needed) rules (e.g. rules 9.23 and 9.24) The problem with if added rules is that they could derive a large number of useless truths. For instance, if rule 9.23 ....

J. Crawford and B. Kuipers. Algernon: a tractable system for knowledge representation. SIGART Bulleting, 2(3):35--44, 1991.

....the case. If the identity of ds # can be established, then s is asserted in pmodel.This declares ds # to be known in pmodel and creates the places and paths that explain s according to the axioms of the topological theory TT#E#. DRAFT November 13, 2001 39 the rule based system Algernon [Crawford and Kuipers, 1991] as our theorem prover. In Section 8.4 (Page 40) we present an illustrative trace of the algorithm. 8.1 Partial Models In addition to a list of schemas to explain, a partial model has associated a set of objects (i.e. distinctive states, schemas, places, paths) that are known in the model. The ....

J. Crawford and B. Kuipers. Algernon: a tractable system for knowledge representation. SIGART Bulleting, 2(3):35--44, 1991.

....these units are referred to as frames ; Each frame encodes a CG along with addition information, in the form of access paths, about how to infer values of particular nodes (see Section 3. 2) This working note describes now conceptual graphs can be mapped onto the representation language Algernon [Crawford and Kuipers, 1991, Crawford, 1990] Section 3 describes one way in which CGs can be encoded in Algernon, where a set of rules encodes the various implications which a CG contains. Section 4 discusses this and alternative ways that CGs could be encoded with Algernon, and their relative merits and weaknesses. 2 ....

Crawford, J. M. and Kuipers, B. J. (1991). Algernon -- a tractable system for knowledge-representation. SIGART Bulletin, 2(3):35--44.

No context found.

Crawford, J. and B. Kuipers. Algernon - A Tractable System for Knowledge-Representation. In working notes: AAAI Spring Symposium on Implemented Knowledge Representation and Reasoning Systems, Palo Alto, CA 1991.

....methods for qualitative reasoning about dynamically changing worlds. Our methods are being applied in a problem solver which addresses textbook problems from the magnetic fields domain. The problem solver makes use of the QPC [3] and QSIM [11] qualitative reasoning systems and the Algernon [4] knowledge representation system. This implementation architecture will allow us to inherit additional capabilities as they are developed for the underlying systems, such as methods for incorporating incomplete quantitative knowledge into the reasoning process [12, 10] This paper describes two ....

....and is asked to produce an integrated knowledge base description that captures the spatial and dynamic state of the world given in the initial scenario. In the current implementation, The idraw graphical editor is used to generate the diagram, and the Algernon knowledge representation language [4] is used to maintain the knowledge base. The first step in integrating diagram and text input, after choosing a frame of reference for the diagram, is to associate a common semantics with the objects in the diagram and the text. For objects in diagrams, we use shading and patterns to designate ....

J. Crawford and B. Kuipers. Algernon - a tractable system for knowledge-representation. In AAAI Spring Symposium on Implemented Knowledge Representation and Reasoning Systems, Palo Alto, CA, 1991.

No context found.

J. Crawford and B. Kuipers. Algernon - A Tractable System for Knowledge-Representation. In Proceedings of the AAAI Spring Symposium on Implemented Knowledge Representation and Reasoning Systems, Palo Alto, CA.

No context found.

Crawford, J. and B. Kuipers. Algernon - A Tractable System for Knowledge-Representation. AAAI Spring Symposium on Implemented Knowledge Representation and Reasoning Systems, Palo Alto, CA 1991.

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