P.J.F. Lucas & L.C. van der Gaag. Principles of Expert Systems. Addison-Wesley: Wokingham, 1991. 11  Home/Search   Document Not in Database   Summary   Related Articles   Check

....elements are deleted from the knowledge base, thus obtaining a personalized knowledge base. This personalized knowledge base is denoted by KN , where N indicates that this knowledge base corresponds to information need N . This plan of attack has been proven effective in expert systems (see e.g. [LG91]) 9 At later stages in the query session, the personalized knowledge base is used instead of the original one. This causes less documents to be proven relevant, exactly those that relied on invalid elements of the knowledge base. In an iterative process, this leads to a suitable knowledge base ....

P.J.F. Lucas and L.C. van der Gaag. Principles of Expert Systems. AddisonWesley, Reading, Massachusetts, 1991. 11

....logic, decision making and knowledge processing to categorise, consult, analyse and diagnose. An expert system uses information which has been borrowed from human experts in a field and stored in the system s knowledge base. This information is thus the result of experience in the problem domain [8]. Answers are automatically extracted from the knowledge base by an inference process which is invisible to the user. For a system to be useful however, it should not only give answers, but should be able to explain the steps it has taken in reaching them, just as a human expert may explain the ....

P. Lucas, and L. Van der Gaag, Principles of Expert Systems, Addison-Wesley Publishing Company, 1991, p. 6.

....be used as a formalism to solve various tasks. A common application of belief network is diagnosis, for example in medicine. Typically, a user enters patient data into a belief network, after which a diagnosis can be determined using a probabilistic inference algorithm (cf. Van der Gaag, 1999, Lucas Van der Gaag, 1991] In the rst exercise, you will experiment with a small, but well designed belief network concerning Wilson s disease. This disorder is characterized by a derangement of the copper metabolism in the human body, due to lack of the copper binding serum protein caeruloplasmin. This causes an ....

P.J.F. Lucas & L.C. van der Gaag (1991). Principles of Expert Systems, Addison-Wesley, Wokingham.

....in dealing with patients who are mechanically ventilated and display symptoms and signs possibly related to the development of pneumonia. In this paper, we describe the design of a decision theoretic model, i.e. a model based on a combination of the theory of (causal) probabilistic networks [23,15,31], also known as Bayesian (belief) networks [37] and decision theory [41] that is aimed at supporting clinicians in prescribing antibiotic therapy to mechanically ventilated patients with pneumonia at the ICU. It is part of a decision theoretic expert system [25,27] called PTA (Pneumonia Therapy ....

....as was mentioned above, is one of the main issues of the problem. It will be the subject of the present section. 4.1. The probabilistic network formalism A natural representation of the uncertainties involved in treating patients with pneumonia is offered by the probabilistic network formalism [31,37]. This formalism allows for the representation of causal relationships, either explicitly indicated as being temporal in nature or not. Such relationships were used in the design of the dynamic model discussed above. Formally, a probabilistic network is an acyclic directed graph G= V(G) A(G) ....

Lucas PJF, Van der Gaag LC. Principles of Expert Systems. Reading, MA: Addison-Wesley, 1991.

....data values. x5 is an example refinement using the Blocks World as a simple system involving search. Finally x6 concludes by describing the performance of the mechanisms, related work and further research. 2 3 Rule Based Transition Systems Knowledge Based Systems (KBS) e.g. Lucas Van Der Gaag [Lucas Van Der Gaag, 1991] is a branch of Artificial Intelligence which is concerned with constructing computer programs which perform some task which is traditionally thought of as requiring some degree of human intelligence. For example, medical diagnosis, computer system configuration, action planning and game playing ....

Lucas P. & Van Der Gaag L. 1991 Principles of Expert Systems. Addison-Wesley. International Computer Science Series.

....rule based systems. This is a modification of non deterministic conditional ground term rewriting systems (Jounanaud Dershowitz, 1990) where we extend terms with some useful features for pattern matching and extend the matching process using equational theories. See (Beynon Davies, 1993) (Lucas 4 Van Der Gaag, 1991) or (Luger Stubblefield, 1989) for a general introduction to rule based systems; see (Goguen, 1976) for an introduction to equational algebra. 2.1 Syntax and semantics Let V and F be two disjoint sets of variables and function symbols respectively. Each function symbol f 2 F has a positive ....

Lucas P. & Van Der Gaag L. (1991) Principles of Expert Systems. Addison-Wesley. International Computer Science Series.

.... The main applications of the formalisms are in classification, e.g. diagnosis (cf. 4] and in decision making under uncertainty, e.g. optimal treatment management of a patient (cf. 1] The decision theoretic network formalisms originate from two different fields: knowledgebased systems [11], and statistical decision theory [16] which is also reflected in the various issues that arise when building decision theoretic expert systems. As with any expert system, extracting knowledge of a specific domain from various sources, such as experts, literature and databases, is required in the ....

Lucas, PJF, Gaag, LC van der, 1991. Principles of Expert Systems, Addison-Wesley, Wokingham. 12

....at the top, the treatment(s) that are unsuitable because of a contra indication are presented in shaded characters in a seperate box (see Figure 4) Treatment decisions 8 The data base The data base contains the decision elements, described above. They are represented in frames (compare Lucas van der Gaag, 1991), which allows for easy change or addition of elements (see below: Expert mode) There is one super frame therapies , with the twelve modalities as sub frames. There is one super frame symptoms , with the four categories of symptoms as sub frames and each symptom as sub frame in the appropriate ....

Lucas, P. & Gaag, L.C. van der (1991). Principles of expert systems. Workingham, UK: Addison Wesley.

....in dealing with patients 2 who are mechanically ventilated and display symptoms and signs possibly related to the development of pneumonia. In this paper, we describe the design of a decision theoretic model, i.e. a model based on a combination of the theory of (causal) probabilistic networks [23, 15, 31], also known as Bayesian (belief) networks [37] and decision theory [41] that is aimed at supporting clinicians in prescribing antibiotic therapy to mechanically ventilated patients with pneumonia at the ICU. It is part of a decision theoretic expert system [25, 27] called PTA (Pneumonia Therapy ....

....as was mentioned above, is one of the main issues of the problem. It will be the subject of the present section. 4. 1 The probabilistic network formalism A natural representation of the uncertainties involved in treating patients with pneumonia is o ered by the probabilistic network formalism [31, 37]. This formalism allows for the representation of causal relationships, either explicitly indicated as being temporal in nature or not. Such relationships were used in the design of the dynamic model discussed above. Formally, a probabilistic network is an acyclic directed graph G = V (G) ....

Lucas PJF, Van der Gaag LC. Principles of Expert Systems. Wokingham: AddisonWesley, 1991.

.... ned as a pair B = G; P ) where G = V (G) A(G) is an acyclic directed graph with a set of vertices (or nodes) V (G) fX 1 ; X 2 ; X n g and a set of arcs A(G) V (G) V (G) and where P is a joint probability distribution de ned on the variables corresponding to the vertices V (G) [4]. The basic property of a Bayesian belief network is that the joint probability distribution P (X 1 ; X 2 ; X n ) is equivalent to the product of the (conditional) probabilities which are speci ed for the network; formally: P (X 1 ; X 2 ; X n ) n Y i=1 P (X i j (X i ) ....

P.J.F. Lucas & L.C. van der Gaag. Principles of Expert Systems. Addison-Wesley: Wokingham, 1991. 11

....in terms of likelihood ratios, yielding mathematically sound probabilistic interpretations of the model [9] A certainty factor CF(h; e) is just a numerical measure between 1 and 1, de ned in terms of measures of belief and disbelief. The actual de nition is not relevant for this paper (cf. [15]) A negative certainty factor indicates that the hypothesis h is discon rmed by the evidence e; a positive certainty factor indicates that the hypothesis h is con rmed by the evidence e. A certainty factor equal to zero indicates that the evidence e does not in uence the belief in the ....

P.J.F. Lucas and L.C. van der Gaag. Principles of Expert Systems (Addison-Wesley, Wokingham, 1991).

....to build, refine and evaluate temporal Bayesian network models. 6.4. 2 Modelling methods for the development of temporal Bayesian models One of the strongest features of the Bayesian network formalism is that it supports representing both qualitative and quantitative knowledge [50, 39, 26, 55, 56, 57]. At a practical level this means that suitably encoded knowledge from clinical experts can be used to guide the search for temporal models in learning, for example by utilising particular fragments of a temporal Bayesian model that are clinically obvious, but may nonetheless be di#cult to learn ....

P.J.F. Lucas and L.C. van der Gaag. Principles of Expert Systems. Wokingham: Addison-Wesley, Wokingham, 1991.

....of likelihood ratios, yielding mathematically sound probabilistic interpretations of the model [9] A certainty factor CF(h,e) is just a numerical measure between 1 and 1, defined in terms of measures of belief and disbelief. The actual definition is not relevant for this paper (cf. Ref. [15]) A negative certainty factor indicates that the hypothesis h is disconfirmed by the evidence e; a positive certainty factor indicates that the hypothesis h is confirmed by the evidence e. A certainty factor equal to zero indicates that the evidence e does not influence the belief in the ....

P.J.F. Lucas, L.C. van der Gaag, Principles of Expert Systems, Addison -Wesley, Wokingham, UK, 1991.

.... G = V (G) A(G) with a set of vertices V (G) fV 1 ; V n g, representing a set of stochastic variables V, and a set of arcs A(G) V (G) V (G) representing conditional and unconditional stochastic independences among the variables, modelled by the absence of arcs among vertices [12]. The basic property of a Bayesian network is that any variable corresponding to a vertex in the graph G is conditionally independent of its non descendants given its parents; this is called the local Markov property. On the variables V is de ned a joint probability distribution Pr(V 1 ; ....

....an independent form Bayesian network, by way of analogy with the special form of Bayes rule, called its independent form, for which the same assumptions hold. The independent form is used to compute the a posteriori probability of a class value c k given the evidence E fe 1 ; e m g [12]: Pr(c k j E) Pr(E j c k ) Pr(c k ) Pr(E) e2E Pr(e j c k ) Pr(c k ) j=1 e2E Pr(e j c j ) Pr(c j ) where class variable C has q mutually exclusive values c k , and Pr(E) 0. The model is normally used for classi cation of cases, i.e. as a classi er. Note that Pr(E j c k ) ....

P.J.F. Lucas and L.C. van der Gaag. Principles of Expert Systems. Addison-Wesley, Wokingham, 1991.

.... Unlike the abductive theory of diagnosis, only a single concept of causality is employed, which is only made more expressive by the interpretation of causal relations as conditional probabilities [Peng 2z Reggia, 1990] yielding a formalism that is much alike the belief network formalism [Lucas 2z Van der Gaag, 1991; Pearl, 1988] Furthermore, a knowledge base consists only of a specification of single defects in terms of associated findings; in this way, it is not possible to model interactions among defects. More over, the notion of set covering diagnosis (explanation) is fixed, with the exception of the ....

P.J.F. Lucas and L.C. van der Gaag (1991). Principles of Ex- pert Systems. Wokingham: Addison-Wesley.

No context found.

P.J.F. Lucas, L.C. van der Gaag. Principles of Expert Systems. Wokingham: Addison-Wesley, 1991.

.... graph G = V (G) A(G) with a set of vertices V (G) fV 1 ; V n g, representing a set of stochastic variables V, and a set of arcs A(G) V (G) V (G) representing conditional and unconditional stochastic independences among the variables, modelled by absence of arcs among vertices [5, 6]. The basic property of a Bayesian network is that any variable corresponding to a vertex in the graph G is (conditionally) independent of its non descendants; this is called the local Markov property. On the variables V is de ned a joint probability distribution Pr(V 1 ; V n ) 2 C E 1 ....

....be called an independent form Bayesian network, by way of analogy with the special form of Bayes rule, called its independent form, for which the same assumptions hold. The independent form of Bayes rule is used to compute the a posteriori probability of a class value c k given the evidence E [6]: Pr(c k j E) Pr(E j c k ) Pr(c k ) Pr(E) Q e2E Pr(e j c k ) Pr(c k ) P q j=1 Q e2E Pr(e j c j ) Pr(c j ) where class variable C has q mutually exclusive values, and Pr(E) 0. Note that Pr(E j c k ) Y e2E Pr(e j c k ) holds, because of the assumption that the evidence ....

P.J.F. Lucas, L.C. van der Gaag. Principles of Expert Systems. Addison-Wesley, Wokingham, 1991.

....in terms of likelihood ratios, yielding mathematically sound probabilistic interpretations of the model [9] A certainty factor CF(h; e) is just a numerical measure between 1 and 1, de ned in terms of measures of belief and disbelief. The actual de nition is not relevant for this paper (cf. [15]) A negative certainty factor indicates that the hypothesis h is discon rmed by the evidence e; a positive certainty factor indicates that the hypothesis h is con rmed by the evidence e. A certainty factor equal to zero indicates that the evidence e does not in uence the belief in the hypothesis ....

P.J.F. Lucas and L.C. van der Gaag, Principles of Expert Systems, Addison-Wesley, Wokingham, UK, 1991.

....able to make explicit. Formalizing such knowledge may yield rules with a wider application than intended. As an example, consider the production rule depicted in Figure 1. In this HEPAR rule we use the object attribute value representation and certainty factors as employed in the DELFI 2 system [9, 13]. This rule was originally formulated without the second condition (indicated by the right arrow) for the modified rule to be applicable, fever must be absent in the patient. So, in its original form it is an example of a too weakly formulated production rule. This missing condition caused the ....

P.J.F. Lucas, L.C. van der Gaag, Principles of Expert Systems (Addison-Wesley, Wokingham, 1991).

....the uncertainty in problem solving knowledge to the final conclusions concerning the goal attributes. The representation of knowledge by means of production rules with object attribute value triples, as well as top down inference and the certainty factor model are described in more detail in [1, 12]. 3 Design of the belief network 3.1 Basic notions of belief networks The formalism of belief networks offers an intuitively appealing approach for expressing inexact causal relationships between domain concepts [7, 20] A belief network consists of two components [3] ffl A qualitative ....

....networks. However, several techniques are known for extending the method of Kim and Pearl to arbitrary belief networks [20] For an in depth account of belief networks, the interested reader is referred to [18, 20] For an overview of the principles of belief networks, the reader is referred to [12]. There are now several expert system shells available for building belief networks. For the experiments described in this article, we used the IDEAL system [27] to which some small programs for the graphical display and printing of belief networks were added. 3.2 Design considerations In ....

P.J.F. Lucas and L.C. van der Gaag, Principles of Expert Systems (Addison-Wesley Publishing Company, Wokingham, England, 1991).

.... graph G = V (G) A(G) with a set of vertices V (G) fV1 ; Vng, representing a set of stochastic variables V, and a set of arcs A(G) V (G) V (G) representing conditional and unconditional stochastic independencies among the variables, modelled by absence of arcs among vertices [5, 6, 9]. On the variables V is de ned a joint probability distribution Pr(V1 ; Vn ) for which the following decomposition property holds: Pr(V1 ; Vn) n Y i=1 Pr(V i j (V i ) where (V i ) denotes the conjunction of variables corresponding to the parents of V i , for i = 1; ....

....with the special form of Bayes rule, called its independent form, for which the same assumptions hold. This form of Bayes rule is also known as the naive Bayes rule. The independent form of Bayes rule is used to compute the a posteriori probability of a class value ck given the evidence E [6]: Pr(ck j E) Pr(E j ck ) Pr(ck ) Pr(E) Q e2E Pr(e j ck ) Pr(ck ) P q j=1 Q e2E Pr(e j c j ) Pr(c j ) where class variable C has q mutually exclusive values, and Pr(E) 0. Note that Pr(E j ck ) Y e2E Pr(e j ck ) holds, because of the assumption that the evidence variables E i ....

P.J.F. Lucas, L.C. van der Gaag. Principles of Expert Systems. Addison-Wesley, Wokingham, 1991.

....causality. Unlike the abductive theory of diagnosis, only a single concept of causality is employed, which is only made more expressive by the interpretation of causal relations as conditional probabilities [Peng Reggia, 1990] yielding a formalism that is much alike the belief network formalism [Lucas Van der Gaag, 1991; Pearl, 1988] Furthermore, a knowledge base consists only of a specification of single defects in terms of associated findings; in this way, it is not possible to model interactions among defects. Moreover, the notion of set covering diagnosis (explanation) is fixed, with the exception of the ....

P.J.F. Lucas and L.C. van der Gaag (1991). Principles of Expert Systems. Wokingham: Addison-Wesley.

....section, we briefly review the syntax and semantics of first order predicate logic. Furthermore, some of the considerations to make logic a practical language for building expert systems are discussed. Only the notions required for understanding the remainder of this paper are discussed here (cf. [11, 24, 41]) 2 2.1 Syntax and semantics First order predicate logic essentially is a language to express knowledge concerning the relationship between individual objects and classes of object. Syntactically, each relation is expressed using a predicate symbol, such as P ; the objects in the relation are ....

....of diagnostic medical knowledge. Diagnostic medical knowledge is represented in the HEPAR system using production rules with object attribute value tuples. According to the declarative reading of rules, translation of most production rules is straightforward yielding logical implications [2, 24]. An example of such a logical implication concerning Wilson s disease is shown below: 8x(Duration(x; complab,chronic) disorder(x) hepatocellular) age(x) 25) caeruloplasmin(x; labresult,biochemistry) 20) urinary copper(x; labresult,biochemistry) 1) Diagnosis(x; Wilson s ....

P.J.F. Lucas, L.C. van der Gaag, Principles of Expert Systems, (Addison-Wesley, Wokingham, 1991). 18

....increasing extent. Especially the area of knowledge based systems has attracted much attention. The phrase knowledge based system, or expert system, is generally employed to denote computer systems in which some symbolic representation of human knowledge is incorporated and applied [Jackson, 1990, Lucas van der Gaag, 1991] Knowledge based systems are typically designed to deal with real life problems that require considerable human knowledge and expertise for their solution; examples range from medical diagnosis and technical trouble shooting to nancial advice and product design. It is their ability to capture ....

P.J.F. Lucas and L.C. van der Gaag. Principles of Expert Systems. Addison-Wesley, Wokingham, 1991.

No context found.

P. Lucas and L. Van Der Gaag. Principles of Expert Systems. Addison-Wesley Publishing Company, 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