C. Pierret-Golbreich and X. Talon. TFL: An algebraic language to specify the dynamic behaviour of knowledge-based systems. The Knowledge Engineering Review, 11(3):253--280, 1996.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....this version because of space limitations. 1 Introduction Over the last few years a number of formal specification languages have been developed for describing knowledge based systems (KBSs) Examples are DESIRE [vLPT92] KARL [Fen95b] K BS SF [SitV94] ML) 2 [vHB92] MLPM [FG96] and TFL [PGT96] In these specification languages one can describe both knowledge about the domain and knowledge about how to use this domain knowledge in order to solve the task which is assigned to the system. On the one hand, these languages enable a specification which abstracts from implementation details: ....

C. Pierret-Golbreich and X. Talon. TFL: An algebraic language to specify the dynamic behaviour of knowledge-based systems. The Knowledge Engineering Review, 11(3):253--280, 1996.

.... systems, dynamics, inference control, update logics 1 Introduction Over the last years a number of formal specification languages have been developed for describing knowledgebased systems (KBSs) Examples are DESIRE [1, 2] KARL [3, 4] K BS SF [5, 6] ML) 2 [7] MLPM [8] and TFL [9]. In these specification languages one can describe both knowledge about the domain and knowledge about how to use this domain knowledge in order to solve the task which is assigned to the system. On the one hand, these languages enable a specification which abstracts from implementation details: ....

....to structure our analysis. In order to give the reader an impression of what specifications of KBSs may look like in the 5 Process Algebra [17] is an exception. In fact, the semantics of LCM (see Section 4.5) is based on Process Algebra. In the area of KBSs, Process Algebra is used in TFL [9]. 2 elementary transition composed transition operationalization semantics syntax state proof systems Figure 1: The two dimensions of our analysis. different approaches, and in order to illustrate some criticisms of the approaches, we will describe an example of a reasoning task in Section 3. ....

C. Pierret-Golbreich and X. Talon, "TFL: An algebraic language to specify the dynamic behaviour of knowledge-based systems," The Knowledge Engineering Review, vol. 11, no. 3, pp. 253--280, 1996.

.... systems, dynamics, inference control, update logics 1 Introduction Over the last years a number of formal specification languages have been developed for describing knowledge based systems (KBSs) Examples are DESIRE [1, 2] KARL [3, 4] K BS SF [5, 6] ML) 2 [7] MLPM [8] and TFL [9]. In these specification languages one can describe both knowledge about the domain and knowledge about how to use this domain knowledge in order to solve the task which is assigned to the system. On the one hand, these languages enable a specification which abstracts from implementation details: ....

....a valid enterprise. However, when comparing the syntax, for instance, this difference has to be kept in mind: specification 5 Process Algebra [17] is an exception. In fact, the semantics of LCM (see Section 4.5) is based on Process Algebra. In the area of KBSs, Process Algebra is used in TFL [9]. 3 Figure 1: The two dimensions of our analysis. languages usually have an extensive syntax (for the benefit of ease for the programmer) whereas logics often have a small (fixed) syntactic vocabulary (making proofs about the logic shorter) The paper is organized as follows. First, in Section ....

C. Pierret-Golbreich and X. Talon, "TFL: An algebraic language to specify the dynamic behaviour of knowledge-based systems," The Knowledge Engineering Review, vol. 11, no. 3, pp. 253--280, 1996.

....Dynamic Database Logic) and the approach of Abstract State Machines. 1 Introduction Over the last years a number of formal specification languages have been developed for describing knowledge based systems (KBSs) Examples are DESIRE [19] KARL [8] K BS SF [27] ML) 2 [15] MLPM [9] and TFL [24]. In these specification languages one can describe both knowledge about the domain and knowledge about how to use this domain knowledge in order to solve the task which is assigned to the system. On the one hand, these languages enable a specification which abstracts from implementation details: ....

C. Pierret-Golbreich and X. Talon. TFL: An algebraic language to specify the dynamic behaviour of knowledge-based systems. The Knowledge Engineering Review, 11(3):253--280, 1996.

....Instead, only some major fragments of the example are presented. 1 Introduction Over the last years a number of formal specification languages have been developed for describing knowledge based systems (KBSs) Examples are DESIRE [18] KARL [8] K BS SF [26] ML) 2 [14] MLPM [9] and TFL [23]. In these specification languages one can describe both knowledge about the domain and knowledge about how to use this domain knowledge in order to solve the task which is assigned to the system. On the one hand, these languages enable a specification which abstracts from implementation details: ....

C. Pierret-Golbreich and X. Talon. TFL: An algebraic language to specify the dynamic behaviour of knowledge-based systems. The Knowledge Engineering Review, 11(3):253--280, 1996.

....the semantics of the different primitives. Completeness implies that the system do not abandon the solving process before possibilities have been tried. Some systems have been formalised using different logical languages or new formal languages: K BS SF[JS92] KARL [Fen95] ML) 2 [HB92] TFL [PT96] [TP97] etc. However, these formalisations are so complex that the behaviour of these systems is difficult to be predicted or proved. MetaTask, a framework dedicated to the control representation. MetaTask has been developed to tackle the control problem in a better way. The two main ....

....and semantics, opposed to other task languages, e.g. LISA [Del93] which are only packages providing specific primitives, kinds of macro defined to capture the control knowledge. The detailed syntax of the MetaTask language and its semantics, defined thanks to ADTs, are presented in [Tal98] [PT96]. We only give here an informal presentation of its primitives specific to the control: Problem, Method, Task Module, and Strategy 1. Problems. MetaTask makes a clear distinction between what is expected to be achieved by the system, and how it should be achieved. The primitive Problem refers to ....

[Article contains additional citation context not shown here]

Pierret-Golbreich, C., Talon, X., "TFL, an algebraic language to specify the dynamic behavior of Knowledge-Based Systems." The Knowledge Engineering Review, 11(3), September 1996.

....specification and human s understanding of the KBS [19] A formal model (FM) encodes knowledge in a symbolic formalism with a mathematically sound basis and a declarative semantics. It allows to eliminate ambiguities and inconsistencies from the CM, and enables formal verification and validation [11, 17]. A design model (DM) is concerned with, among others, an efficient realisation of the CM and or FM on a computer. It specifies, among others, the data structures, the algorithms and the architecture of the target application. A DM also records the rationale for the choices made [15, 23] 3.2. ....

....expressiveness such as (ML) 2 [25] otherwise there is the risk that we cannot represent all knowledge in the conceptual model. On the other hand, if our aim is validation, then an operational FM such as KARL [12] would be more appropriate (limiting the expressive power of the FM) See also [13, 17, 11]. Analogously, we can have different types of DMs [23, 15] 4. Supporting the transition paths Past experience has shown that good methodologies will not work in practice if they are not properly supported with software tools and guidelines. This holds also for current KE methodologies. ....

C. Pierret-Golbreich and X. Talon. Tfl: algebraic language to specify the dynamic behaviour of knowledge-based systems. The Knowledge Engineering Review, 11(2), 1996.

....decomposition considering that , #and# #in the figure 3 can represent an atomic positive or negative proference or any complex preference. 8sq. i### #q ersgev#xy e syx#py # ipi ixgif 2 Semantics of preferences We specify the preferences in the formal framework of Abstract Data Types (see [PT96] for details) by the following sort ipi ixgi : fy # ipi ixgi . fi#fyyviex##wi ryh##fi Ywi ryha yi e syxf # # ipi ixgi iw o oo# ipi ixgi # # # ipi ixgi# # ipi ixgi# #fyyviex# fivyxqf# y e ywsg# # ipi ixgi# #fyyviex# sf#e ywsg ## #wi ryh# # ipi ixgi xiqe s i# ipi ixgi ....

....irrelevant at another step. To capture such context dependency, we introduce conditional strategies of the form Q# spp# gyxh which means that the strategy Q# is defined when the condition gyxh# is true regarding to the context 5# (gyxh#is expressed in terms of boolean expressions, cf. [PT96]) Graphically, the condition is represented by including the condition at the top of the frame delimiting the conditioned strategy (see on the right) 2 Strategy evaluation At a particular state of the resolution the preference which results of a strategy depends on the set of methods which ....

Pierret-Golbreich, C., Talon, X., "TFL, an algebraic language to specify the dynamic behavior of Knowledge-Based Systems." The Knowledge Engineering Review, 11(3), September 1996.

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