Expert systems differences

Expert systems differ from standard procedural or object-oriented programs in that there is no clear order in which code executes. Instead, the knowledge of the expert is captured in a set of rules, each of which encodes a small piece of the expert s knowledge. Each rule has a left-hand side and a right-hand side. 

It extends the usage of statistical methods and combines it with machine learning methods and the application of expert systems. The visualization of the results of data mining is an important task as it facilitates an interpretation of the results. Figure 9-32 plots the different disciplines which contribute to data mining. 

In the first stages of the development of an Action plan all control options are considered. In the case of lakes, this process is aided by a PC-based expert system , PACGAP, which looks at the physical and chemical characteristics of the lake to determine the most likely option for control. Once further, more detailed information has been collected on the lake s nutrient inputs and other controlling factors, amore complex interactive model can be used (Phytoplankton Response To Environmental CHange, PROTECH-2) to define the efficacy of proposed control options more accurately. This model is able to predict the development of phytoplankton species populations under different nutrient and stratification regimes. 

A Russian expert system, PASS (prediction of activity spectra for substances) [84], uses substructural descriptors called multilevel neighborhoods of atoms [85] to predict over 900 different pharmacological activities from molecular structure. These activities include a number of toxicity end points such as carcinogenicity, mutagenicity, teratogenicity, and embryotoxicity. The accuracy of prediction has been shown [86] to range from about 85% to over 90%. One-off predictions can be obtained free of charge on the PASS website [84]. 

The most popular representation scheme in expert systems is the mle-based scheme. In a rule-based system the knowledge consists of a number of variables, also called attributes, to which a number of possible values are assigned. The rules are the functions that relate the different attributes with each other. A mle base consists of a number of If... Then... mles. The IF part contains the conditions that must be satisfied for the actions or conclusions in the THEN part to be valid. As an example, suppose we want to express in a mle that a compound is unstable in an alkaline solution if it contains an ester function. In a semi-formal way the mle can be written ... 

Rules seemingly have the same format as IF.. THEN.. statements in any other conventional computer language. The major difference is that the latter statements are constructed to be executed sequentially and always in the same order, whereas expert system rules are meant as little independent pieces of knowledge. It is the task of the inference engine to recognize the applicable rules. This may be different in different situations. There is no preset order in which the rules must be executed. Clarity of the rule base is an essential characteristic because it must be possible to control and follow the system on reasoning errors. The structuring of rules into rule sets favours comprehensibility and allows a more efficient consultation of the system. Because of the natural resemblance to real expertise, rule-based expert systems are the most popular. Many of the earlier developed systems are pure rule-based systems. 

Because the rules or procedures in expert systems are heuristic they are often not well-defined in a logical sense. Nevertheless, they are used to draw conclusions. A conclusion can be uncertain because the truth of the rules deriving it cannot be established with 100% certainty or because the facts or evidence on which the rule is based are uncertain. Some measure of reliability of the obtained conclusions is therefore useful. There are different approaches used in expert systems to model uncertainty. They can be divided into methods that are based on... 

The appearance of expert systems to solve practical problems, also in chemistry, started in the eighties. During this period much experience has been acquired through the expected and unexpected problems that arose during such projects. Until now there are only a few commercially available expert systems and this is not likely to change in the near future. This implies that expert systems will be mostly in-house developments. The different steps to consider are ... 

The next few steps are very similar to those required in any software project. One of the first stages is the clear definition of the knowledge domain. It must be clear which problems the expert system must solve. It is at this stage not the intention to define how this can be done. Clarity and specificity must be the major guides here. Fuzziness at this stage will, more than in classical software projects, have to be paid for later when different interpretations cause misunderstandings. Equally important is the clear definition of the end user(s). An expert system set up as decision support tool for professionals is totally different from an expert system that can be used as a training support for less professional people. 

The testing phase is important in expert system development. The practical applicability of the expert system will largely depend on this phase. Testing expert systems is different from normal software engineering in a number of ways. First, it is difficult to test exhaustively the full code and all possible paths the reasoning process may follow. Secondly, the nature of expert systems poses some typical problems. Due to their heuristic nature the correctness of the results cannot be easily verified. A certain degree of errors may be acceptable and, moreover, an... 

An early field of application in analytical chemistry is structure elucidation. DENDRAL was one of the first ES in general, designed to the identification of organic compounds from mass spectrometric data (Buchanan and Feigenbaum [1978]). In the 1980s and 1990s a flood of expert systems has been developed in analytical chemistry for different types of application, viz ... 

An expert system does much more than extract information from a database, format it, and offer it up to the user it analyzes and processes the information to make deductions and generate recommendations. Because an ES may be required to present alternative strategies and give an estimate of the potential value of different courses of action, it must contain a reasoning capacity, which relies on some sort of general problem-solving method. 

In chlor-alkali production, EMOS should be able to determine problems with both anode coatings and membranes. The literature is replete with examples of the effect of different impurities on membranes [2] and of the analysis of different problems using polarisation curves to determine their cause [3, 4]. These analysis techniques have been incorporated into the expert system in the form of approximations of the polarisation curves. Use is made of the familiar k-factor (see Equation 8.2) or the more accurate logarithmic form of this factor (Equation 8.3) ... 

The primary differences, then, between development of expert systems and more traditional software engineering are found in steps one and two, above. First, the problems chosen will involve symbolic reasoning, and will require the transfer of expertise from experts to a knowledge base. Second, rapid prototyping, the "try it and see how it works, then fix it or throw it away" approach will play an important role in system development. 

Explanation. A user may ask for explanation of the line of reasoning at any time during an expert system consultation. RuleMaster presents explanation as a list of premises and conclusions in English-like text. The explanation describes the execution path which led up to the current conclusion or question. Explanation is presented in proof ordering, which usually differs from the order in which the questions and conclusions were encountered. This is perceived as more relevant and understandable than the time-ordered presentation of fired rules, as is present in most expert system approaches. 

RuleMaker, a subsystem of RuleMaster, induces rules for all situations from examples that may cover only some of the cases. At the heart of the induction process is the creation of an induction file, which in part includes examples indicating what the expert system should do under different circumstances. Now, in the example above, THE RULES FOR CORRELATING VARIOUS CHEMICAL AND PHYSICAL PARAMETERS OF THE HAZARDOUS CHEMICALS TESTED WITH THE PROTECTIVE ABILITY OF THE SELECTED GLOVE MATERIALS ARE NOT KNOWN — THEY WILL HAVE TO BE INDUCED FR04 THE ANALYTICAL DATA. 

---