DENDRAL

Two systems will be introduced below on the one hand the DENDRAL system for automatic structure elucidation [25], which was one of the first expert systems on the other hand the EROS system [26], which can be used for simulating reactions. 

DENDRAL followed a three-stage procedure. In the first phase, the so-called plan, prior knowledge, and heuristics were used to deduce a set of constraints. Constraints could be, for example, the exemption of large sets of candidate solutions or the suggestion for a extensive search over limited classes of solutions. 

The second step, the so called generation, created only those structures which complied with the given constraints, and imposed additional restrictions on the compounds such as the number of rings or double bonds. The third and final phase, the tester phase, examined each proposed solution for each proposed compound a mass spectrum was predicted which was then compared with the actual data of the compound. The possible solutions were then ranked depending on the deviation between the observed and the predicted mass spectra. 

A large amount of work was devoted to the development of the DENDRAL system and many highly competent people contributed to this project. In the end, many reasons may have worked together to explain why this project was finally 

The EROS (Elaboration of Reactions for Organic Synthesis) system [26] is a knowledge-based system which was created for the simulation of organic reactions. Given a certain set of starting materials, EROS investigates the potential reaction pathways. It produces sequences of simultaneous and consecutive reactions and attempts to predict the products that will be obtained in those reactions. 

---

The DENDRAL project initiated in 1964 at Stanford was the prototypical application of artificial intelligence techniques - or what was understood at that time under this name - to chemical problems. Chemical structure generators were developed and information from mass spectra was used to prune the chemical graphs in order to derive the chemical structure associated with a certain mass spectrum. 

DENDRAL proved to be fundamentally important in demonstrating how rule-based reasoning could be developed into powerful knowledge-engineering tools [93]. 

R. K. Lindsay, B. G. Buchanan, E. A. Feigenbaum, J. Lederberg, Applications of Artificial Intelligence for Organic Chemistry - The DENDRAL Project, McGraw-Hill, New York, 1980. 

Chemistry — The DENDRAL Project, McGraw-Hill, New York, 1980. 

J. Lederberg, How DENDRAL was conceived and bom, in ACM Symposium on the History of Medical Informatics, National Library of Medidne, 1987. Later published in A History of Medical Informatics, B. I. Blum, K. Duncan (Eds.), Association for Computing, Machinery Press, New York 1990, 14-44. 

B.G. Buchanan and E.A. Feigenbaum, Dendral and Meta-Dendral their application dimension. Artif. Intell., 11 (1978) 5-24. 

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 ... 

Buchanan BG, Feigenbaum EA (1978) DENDRAL and Meta-DENDRAL their applications dimension. Artific Intell 11 5... 

Although most expert systems are designed to advise fairly unsophisticated users, some act instead as assistants to established human experts DENDRAL, which we introduce shortly, falls in this category. Rather than replacing the expert, these systems enhance their productivity. Assistants of this sort have particular value in fields, such as law or medicine, in which the amount of potentially relevant background material may be so large that not even a human expert may fully be cognizant with it. 

Rather than a single program, DENDRAL was a group of interrelated programs that took turns to participate in a discussion with the user, proposing possible molecular structures for the user to consider and removing from consideration any structure that the user deemed not to be chemically stable, incompatible with the observed fragmentation pattern, or otherwise... 

Like all early expert systems, DENDRAL and SHRDLU required exact knowledge to function. The way that expert systems work depends on whether the knowledge that they manipulate is exact ("The temperature is 86°C") or vague ("The temperature is high"). We shall first consider how an ES can use exact knowledge to provide advice. Methods for dealing with ill-defined information form the topic for the next chapter, which covers fuzzy logic. 

According to the Dendral approach, one also needs restrictions for the unknown chemical structure, usually given as substructures that are... 

The well-known DENDRAL and META-DENDRAL programs (43) are noted as the major AI success in chemical applications over the past decade. However, advances in analytical technology and computer capabilities have led to new approaches (44-56). Information fusion from selected instrumental tools often is a more productive route than exhaustive data analysis from a single source. Furthermore, combination of chromatographic separation with spectral, thermal, and microchemical analyses can be realistically achieved in many laboratories. Generalizing and documenting this trend using an AI approach seemed appropriate at this time. 

The DENDRAL project has been reviewed by Gray " and responded to by Feigenbaum and co-workers. 

Searches through more than one data base in combination would be very desirable. For example, one often possesses mass spectral and nmr data for an unknown and it would be very useful to be able to identify any compounds that match these data in a single search. Work is going on in this area do interface programs so that this approach can be tested. In another development, it is expiected that the CONGEN programs developed for the DENDRAL project [30] will be merged into CIS within the... 

---