Chandrasekaran, B., Towards a functional architecture for intelligence based on generic information processing tasks. In: Proceedings of the 10th IJCAI, Milano, IT, 1987, pp. 1183-1192.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....As we have seen, the same type of request may lead to completely different ways of finding the answer. It is dependent on the structure of the legal knowledge, and on the availability of problem solving methods. These two factors are not independent, as the interaction hypothesis suggests [Chandrasekaran, 1987]. The interaction hypothesis states that the structure of knowledge is geared to, or allows specific problem solving methods. A problem solving method is either a decomposition of a problem solving task, or a procedure (algorithm) to solve a problem [Steels, 1990] A problem solving method ....

Chandrasekaran, B., Towards a functional architecture for intelligence based on generic information processing tasks. In: Proceedings of the 10th IJCAI, Milano, IT, 1987, pp. 1183-1192.

....JRC Ispra, Queen Mary Westfield College, Framentec Cognitech, IRIDIA, Iberdrola, Labein, EA Technology, Amber, National Technical University of Athens (NTUA) University of Amsterdam, Volmac, CERN and University of Porto. G. P. Lekkas was with the NTUA and the JRC Ispra. He is now with Praxis Ltd. 20127 Ispra (VA) Italy. N.M. Avouris was with the JRC Ispra. He is now with the University of Patras, Department of Electrical Engineering, 26500 Rio Patras, Greece. G. K. Papakonstantinou is with the NTUA, Computer Science Division, 15773 Athens, Greece. IEEE Log Number 9406077. used [26] ....

....of Athens (NTUA) University of Amsterdam, Volmac, CERN and University of Porto. G. P. Lekkas was with the NTUA and the JRC Ispra. He is now with Praxis Ltd. 20127 Ispra (VA) Italy. N.M. Avouris was with the JRC Ispra. He is now with the University of Patras, Department of Electrical Engineering, 26500 Rio Patras, Greece. G. K. Papakonstantinou is with the NTUA, Computer Science Division, 15773 Athens, Greece. IEEE Log Number 9406077. used [26] Particularly promising to this area of applications is the field of Distributed Artificial Intelligence (DAI) which studies problem solving by ....

[Article contains additional citation context not shown here]

B. Chandrasekaran, "Towards a functional architecture for intelligence based on generic information processing tasks," in Proc. oflJCAl-87, Milan, pp. 1183-1192.

....when several developers are involved or when long term maintenance of a knowledge base is planned to be carried out by prospective users. Current work in the field of expert systems has heavily addressed the notion of taskspecific problem solving methods and architectures, e.g. generic tasks (Chandrasekaran, 1987), and of high level specifications of knowledge based systems, e.g. KADS (Wielinga Schreiber, 1990) see also (Steels, 1990) However, little attention has been given so far to the problem of retrieving relevant facts in large scale domainspecific knowledge bases. With knowledge bases ....

Chandrasekaran, B. (1987). Towards a functional architecture for intelligence based on generic information processing tasks. Proc. 10th Int. Joint Conf. on Artificial Intelligence (IJCAI-87).

....and to schedule them; by optimizing the perception tasks, we mean to compute their parameters, to run the modules and to control their execution. So two main kinds of reasonings (planning and control of execution) have to be performed in the sense of generic tasks introduced by Chandrasekaran [Cha87]; for each of these reasonings, several techniques are available. 2.1 Planning for supervision In this section, the question is: which planning technique is adequate for supervision of perception tasks In problem solving, planning is to find a plan for achieving a goal according to a given ....

B. Chandrasekaran. Towards a Functional Architecture for Intelligence Based on Generic Information Processing Tasks. In Proc. of the 10 th IJCAI, pages 1183--1191, 1987.

....From this proposal stem a group of approaches that proposed using knowledge level models of the reasoning process in the analysis and engineering of AI systems, defining the area we presently call knowledge modeling. Approaches to knowledge level analysis and modeling include Generic Tasks [5], Role limiting Methods [18] Components of Expertise [27] and KADS [25] Their common trace is to identify of a set of high level knowledge primitives from which problem solving knowledge is built. Based on this view about what problem solving is (made of) they propose methods, languages and ....

B. Chandrasekaran. Towards a functional architecture for intelligence based on generic information processing tasks. Planning Agents SIGART Bulletin, Vol. 6, No. 1 In Proceedings of the 10th IJCAI, pages 1183--1192, Milano, 1987.

....relations. Marr [Marr 1982] too emphasized the need to separate the computational theory embodied in an intelligent system from its implementation. He advocated an analysis of information processing in terms of tasks and subtasks, and the mechanisms for accomplishing the tasks. Chandrasekaran [Chandrasekaran 1987] proposed that information processing tasks fall into major classes called generic tasks. All tasks that are instances of the same generic task can be solved by the same methods. This led him to the use of task structures task method subtask decomposition as a way of building AI systems and ....

Chandrasekaran, B., (1987) Towards a functional architecture for intelligence based on generic information processing tasks, In Proc. of the Tenth IJCAI, 1183-1192, Milan.

....steps gives us inference structures [Clancey 1985] and problem solving methods [McDermott 1988] 3. A specification of the tasks performed during problem solving. Whereas it is quite clear that the tasks differ from application to application, researchers try to find abstract task descriptions [Chandrasekaran 1986, 1987, 1988] These three achievements are not to be seen as separating items, but rather as methods for knowledge based systems operating on different levels. The integration of these levels can be seen in the KADS system [Hayward et al. 1987, Schreiber et al. 1988] KADS is a system distinguishing ....

....It is able to modify medical strategies. Generic decision making schemata are tailored to suit the needs of the tasks to be performed during a consultation [Glowinski et al. 1989] 2 A thorough discussion of factors being related to the depth of expert systems can be found in [Bylander 1987, Chandrasekaran et al. 1987, Klein and Finin 1987, Keravnou and Washbrook 1989, Steels 1988, Washbrook and Keravnou 1990] A second form of relation is the functional relation. It describes the function or behavior of components by giving the function of its subcomponents. More and more detailed description levels can be ....

[Article contains additional citation context not shown here]

Chandrasekaran B. (1987): Towards a Functional Architecture for Intelligence Based on Generic Information Processing Tasks, in Proceedings of the 10th International Joint Conference on Artificial Intelligence (IJCAI-87), Morgan Kaufmann, Los Altos, CA.

....communication. At the same time, however, the formal models 1 We will not devote any space to describing the KADS methodology. See other contributions to this issue. 2 This model based approached to knowledge engineering is shared among a number of competing approaches, notably Generic Tasks (Chandrasekaran, 1987), Method to Task (Musen, 1989) and Role Limiting Method (McDermott, 1989) See (Karbach, Vo , Schukey and Drouwen, 1990) for a useful comparison of all these approaches. will be less accessible to domain experts because of their technical complexity, and can therefore not entirely replace the ....

Chandrasekaran, B. (1987). Towards a functional architecture for intelligence based on generic information processing tasks. In Proceedings of the 10th IJCAI, pages 1183--1192, Milano.

....[9, 10] in Common Lisp. In this prototype debugging system, a Generic Task approach is taken [4] The knowledge is arranged based on three tasks, hierarchical classification, routine recognition and data inferencing. The error causes (hypotheses) are arranged in a classification hierarchy [5, 6] where each possible error is a node in the hierarchy. The root node, syntax error , indicates that an error arose. The root has children which make up all of the possible syntax error messages (in this system, there are 50 of these nodes. Each of these errors have several possible causes, which ....

B. Chandrasekaran. Towards a functional architecture for intelligence based on generic information processing tasks. In Proceedings of the Tenth International Joint Conference on Artificial Intelligence, 1987.

....rely on a more principled framework than a mere collection of frames and rules. For instance, heuristic classification models perform their reasoning in terms of general concepts such as Motta p. 7 data , solutions , abstraction and heuristic step . Generic task models, as advocated in [16] are also based on a similar approach. Modelling problem solving in terms of generic tasks has a number of advantages. These models provide principled explanations [17] based on the generic control structure of the problem, and they are more flexible and general. Generic task models can also be ....

Chandrasekaran, B. Towards a Functional Architecture for Intelligence Based on Generic Information Processing Tasks. Proceedings IJCAI 87, Milan, 1987.

....of when it is appropriate. 6 The TIPS Architecture for Problem Solver Integration The experience with Peirce led to investigations on how broadly applicable a sponsor selector system might be as a control structure. The motivation for this work comes from experience with the generic task (GT) [6, 7] approach to system building. Briefly, the GT view emphasizes the identification of domain independent, generic approaches to solving particular kinds of high level problems (e.g. classification, routine design) Once a task is identified, the problem is to identify forms of knowledge for both ....

B. Chandrasekaran. Towards a functional architecture for intelligence based on generic information processing tasks. In Proceedings of the International Joint Conference on Artificial Intelligence 87, pages 1183--1191. International Joint Conference on Artificial Intelligence, 1987.

....elements of the specification. The example in fact is also a specialization of a generic task model for diagnosis. It is possible to seperate the generic task structure from the specific domain, or one could define a generic structure from other sources, e.g. from some libraries addressed by [Chandrasekaran 1987] or [Breuker et. al 1987] The use of having such a generic task is that some kind of a skeleton system already exists which uses units and terminology adopted to the task and which may already incorporate reasoning structure etc. This structure with its task specific building blocks may serve as ....

Chandrasekaran, B. (1987). "Towards a Functional Architecture for Intelligence Based on Generic Information Processing Tasks". In Proc. of the 10th International Joint Conference on Artificial Intelligence, pp.

....feature based search. In the development of most AI systems, the issue of how to represent the knowledge required by the system must be addressed [14] Most knowledge representations are tailored to emphasise those aspects of the domain that are most useful for the particular system s application [29]. Accordingly, the knowledge representation developed in this chapter is tailored to the task of case retrieval. The representation requirements are clarified in section 2.1. In section 2.2 it is argued that a network representation is the most appropriate for the proposed memory model. The main ....

B Chandrasekaran. Towards a Functional Architecture for Intelligence Based on Generic Information Processing Tasks. In IJCAI-87 : Proceedings of the Tenth International Joint Conference on Artificial Intelligence, pages pp1183--1192, Milan, August 1987. Morgan Kaufmann Publishers, Inc.

....concepts and relations, we refer to CGs. 1.2. Problems of Previous Approaches to Conceptual Modelling The problem solving method and the domain structure cannot be modified in systems which follow the role limiting method with several shell editor combinations [Marcus, 1988] Generic tasks [Chandrasekaran, 1987] raise the same problem. Neither the inference level, nor any higher levels can be modified and even the domain structure is fixed. Musens [Musen, 1989a] methodto task approach realised in PROT G allows to generate new tools following the skeletal planrefinement method. These tools incorporate a ....

....ONCOCIN [Musen, 1989b] This may be good for building application oriented tools, where the expert fills his knowledge in forms, but it is inconvenient as a general representation formalism for domain knowledge. The approaches mentioned above are confronted with similar restrictions [Marcus, 1988, Chandrasekaran, 1987]. a fixed problem solving method The role limiting method and generic tasks provide fixed problem solving methods. All possible systems, that can be generated with PROT G are restricted to analytic problems. We want to avoid any restriction to a problem class. KADS] on) CG] is more general ....

Chandrasekaran, B. (1987) Towards a Functional Architecture for Intelligence Based on Generic Information Processing Tasks. In Proc. of the 10th International Joint Conference on Artificial Intelligence, Milano, 1987, pp. 1183-1192.

....the programmer knows what generally may cause each syntax error and knows of ways to determine the specific instruction which caused the error. One approach to diagnostic problem solving is to combine the Generic Tasks of Hierarchical Classification, Routine Recognition and Data Inferencing [3, 4, 6, 7]. These tasks are described in detail elsewhere [1, 2, 5] Here, the tasks will be described as they are used to perform debugging. Hierarchical Classification Hierarchical Classification [2] is a task used for efficient search of a taxonomic hierarchy for conceptual objects (such as hypotheses) ....

B. Chandrasekaran. Towards a functional architecture for intelligence based on generic information processing tasks. In Proceedings of the Tenth International Joint Conference on Artificial Intelligence, 1987.

....Society of Mind program. This is quite a humorous turn of events. The psychologists offer an implemented information processing model, which the computer scientists reject in favor of a creative ############### 3 However, there are responsible practitioners who do care about tractability (Chandrasekaran, 1987). 7 psychological theory The remainder of the epilogue is, essentially, a criticism of certain aspects of modern connectionism, such as distributed representations (Hinton, 1984) from the perspective, and in the diffuse style, of Minsky s vast theory. Conclusion Perceptrons, and its ....

CHANDRASEKARAN, B. (1987). Towards a functional architecture for intelligence based on generic information processing tasks. In Proceedings of the Tenth International Joint Conference on Artificial Intelligence. Milan, 1183-1192.

....architecture in that there is a correspondence between architectures and programming languages. Newell s Knowledge vs. Symbol level level as syntactic issues in the programming language corresponding to the underlying architecture. 2 . 7 GT s as Functional Building Blocks: The Platonic View In [14], Chandrasekaran outlined a view of GT s as building blocks of intelligent systems. We can call it the Platonic View since it proposed that a set of abstract GT s existed which together functionally captured the capability of intelligence in solving problems. The proposal was explicitly not ....

Chandrasekaran, B.: Towards a Functional Architecture for Intelligence Based on Generic Information Processing Tasks. Proceedings of the Tenth International Joint Conference on Artificial Intelligence, (ed. McDermott, J.), 1183-1192, Morgan Kaufmann Publishers, Inc., Los Altos, California, 1987

No context found.

CHANDRASEKARAN, B.Towards a Functional Architecture for Intelligence Based on Generic Information Processing Tasks., IJCAI Vol 2, 1987.

No context found.

Chandrasekaran B. (2) Towards a functional architecture for intelligence based on generic information processing tasks. Proc. of the Tenth IJCAI, Milan (Italy), Vol. 2, 1987, pp 1183-1192.

No context found.

Cha87. B. Chandrasekaran. "Towards a Functional Architecture for Intelligence Based on Generic Information Processing Tasks". In Proc. of the 10th International Joint Conference on Artificial Intelligence, pp. 1183-1192, Milano, 1987.

No context found.

Chandrasekaran, B. Towards a Functional Architecture for Intelligence Based on Generic Information Processing Tasks. Proceedings IJCAI 87, MIlan, 1987.

No context found.

Chan87. Chandrasekaran, B. Towards a Functional Architecture for Intelligence Based on Generic Information Processing Tasks. In Proc. of the 10th International Joint Conference on Artificial Intelligence, Milano, 1987, pp. 1183-1192.

No context found.

Chandrasekaran B. (1987): "Towards a Functional Architecture for Intelligence Based on Generic Information Processing Tasks". Proc IJCAI-87. 1183-1192. Morgan Kaufmann, 1987.

No context found.

Chandrasekaran, B. 1987. Towards a functional architecture for intelligence based on generic information processing tasks, In Proc. of the Tenth IJCAI, 1183-1192, Milan.

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