Data Integration

Abstract

In many problems, we need to classify items, that is, we need to predict some characteristic of an item based on several parameters of the item. Classifications are usually done by classifiers. After being trained on some sample data, these classifiers can be used to classify new data. Section 17.1 describes decision trees, which were frequently used in the nineties by artificial intelligence experts to classify data because they can be easily implemented, and they provide an explanation of the classification. Section 17.2 describes support vector machines, which are increasingly popular classification tools because of the high accuracy of their classifications.

Many challenging applications also require the reuse of the old classifiers to derive a new classifier. Section 17.3 describes the classification integration problem, while Section 17.4 describes the reclassification problem. Section 17.5 discusses data integration via wrappers. Section 17.6 summarizes data fusion. Finally, Section 17.7 discusses the trust management approach to computer security. Constraint databases are helpful in solving all of these problems.