Rule-based classification

In addition, there have also been a number of efforts to find more efficient ways to categorise and enumerate compounds. One approach to defining the relationship of compounds to each other uses a chemical space defined by a set of 500 compounds described with 60 molecular descriptors onto which novel compounds can then be mapped.116 Another approach that has been described uses a rule-based classification scheme to define the common scaffold or core of compounds.117 This approach has been taken further to develop a scoring function to allow the prioritisation of compound libraries for testing.118... 

Schuffenhauer A, Brown N, Ertl P, Jenkins JL, Selzer P, Hamon J. Clustering and rule-based classifications of chemical structures evaluated in the biological activity space. Chem. Inf. Model. 47. 2007 47 325-336. 

Schuffenhauer A, Varin T. Rule-based classification of chemical structures by scaffold. Mol Inform 2011 30 646-664. 

Hybrid systems. Depending on the problem to be solved, use can also be made of a combination of techniques leading to a hybrid system. For example, a rule-based system may use neural networks for solving classification subproblems (as is described in [Hopgood, 1993]), or a combination of a rule-based and a CBR system can be used as in the system for URS data interpretation described later in this paper. 

These selection rules based on their classification as aromatic or antiaronatic transition state are given in the following table. 

FIGURE 5.4 Linear discriminant scores dj for group j by the Bayesian classification rule based on (Equation 5.2). mj, mean vector of all objects in group j Sp1, inverse of the pooled covariance matrix (Equation 5.3) x, object vector (to be classified) defined by m variables Pj, prior probability of group j. 

The aim of supervised classification is to create rules based on a set of training samples belonging to a priori known classes. Then the resulting rules are used to classify new samples in none, one, or several of the classes. Supervised pattern recognition methods can be classified as parametric or nonparametric and linear or nonlinear. The term parametric means that the method makes an assumption about the distribution of the data, for instance, a Gaussian distribution. Frequently used parametric methods are EDA, QDA, PLSDA, and SIMCA. On the contrary, kNN and CART make no assumption about the distribution of the data, so these procedures are considered as nonparametric. Another distinction between the classification techniques concerns the... 

Data of different units and uncertainties can be combined using IF... THEN... rules, based on expert knowledge. Recently, the application of the traceability concept on ecotoxicological studies has been described (Ahlf and Heise, 2007). A suggestion for an ecotoxicological classification system for sediments based on fuzzy sets and fuzzy expert systems is under development (see Chapter 6.2). 

In addition to the two major categories of LEVS approaches, similarity searching and compound classification, database filtering techniques should also be added to the virtual screening spectrum. Rule-based filters that screen databases for compounds with desired (e.g., drug- or leadlike) or undesired (toxicity and instability) molecular properties are widely used, not to identify individual active compounds but rather to reduce the size of screening databases as much as possible for further studies using more sophisticated methods. 

---