Expert systems: data mining

Validation of expert systems still poses a problem that remains to be solved. This is mainly due to the dynamic nature of the reasoning process and the multiplicity of cases that can be solved by an expert system. Whereas the functionality of a linear software system provides the basis for test cases, an expert system can barely be covered comprehensively by a few hundred test cases. Expert systems, as well as most AI software, have to be transitioned carefully from the implementation to the production phase and require a continuous quality control through experimentation as part of the standard maintenance procedure. However, the same applies to a series of other systems, such as knowledge management systems, data mining systems, and, finally, human experts and is no reason for a general denial of expert systems in productive environments. 

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. 

Machine Learning ------M Data Mining [<-------- Expert Systems... 

NLP systems are being developed to address the increasingly challenging problem of data mining for systems level content from the published literature, that is, integrating across the global expert database of biomedical research [71]. One recent approach to this problem was to develop a web-based tool, PubNet, that is able to visualize concept and theme networks derived from the PubMed literature [72]. 

Developments in mass spectrometry technology, together with the availability of extensive DNA and protein sequence databases and software tools for data mining, has made possible rapid and sensitive mass spectrometry-based procedures for protein identification. Two basic types of mass spectrometers are commonly used for this purpose Matrix-assisted laser desorption/ionization (MALDI)-time-of-flight (TOF) mass spectrometry (MS) and electrospray ionization (ESI)-MS. MALDI-TOF instruments are now quite common in biochemistry laboratories and are very simple to use, requiring no special training. ESI instruments, usually coupled to capillary/nanoLC systems, are more complex and require expert operators. We will therefore focus on the use of MALDI-... 

Data mining is an extension to linear searches, mainly based on pattern recognition in a large data set, and is of particular interest in combination with expert systems capabilities in scientific areas. 

The modern way to get rules into an Expert System is not by human experts, but by data mining. 

Heuristic, empirical models can be useful for disseminating the results of corrosion research, as long as the applications are within the range of the data and boundary conditions that the model is built on. These approaches include expert systems and data mining modeling (including semi-empirical, statistical, pattern recognition, neural networks, etc.) models. 

The center of this technology is an expert system for crude assay generation. This expert system, which includes a set of mathematical algorithms based in powerful data-mining techniques and first-principle models, may generate a complete crude oil evaluation from variable levels of information available. 

It provides tools for data analysis. This usually includes the capability of showing the fourier transform or the cross correlation of selected data. More sophisticated information systems contain expert systems or data mining tools, to automatically detect relationships between data. 

Such questions are difficult to answer, as several hundred machine settings and raw material parameters have to be taken into account as potential influencing factors. Data mining technologies based on neuronal networks, expert systems or algebraic methods are used to find the requested answers. Algebraic methods are in many cases not sufficient, as these can deal only with numbers (process values) but not with cardinal data Hke felt supplier, ash suppKer, etc. which are also of high importance. 

The extent of lOS use (e,g, data mining/warehousing, OLAP, DSS, expert systems) among supply chain partners for... 

---