Knowledge Discovery and Data-Mining is a research-oriented course offered primarily to graduate students, and open to undergraduates near the end of their studies, and who have already taken related elective courses. The course will introduce varied aspects of this field, and offer an opportunity for an in-depth understanding of one or more of these aspects, through a hands-on project.
Knowledge discovery in databases (KDD) is a young research and application field, aimed at discovering (and applying) useful knowledge from the large amounts of data existing on computing systems. Such data may be either in the form of (possibly huge) databases, as well as heterogeneous information residing in a haphazard, distributed manner over the internet.
Even when the data resides within a single database, with known attribute formats and semantics, obtaining USEFUL predictive or explanatory knowledge from the data is an open problem. The problem is that the user may not know what queries may be relevant, may not know how it may be relevant, etc. (A typical application question, given a database bank loan applications and returns, might be to find a decision procedure for maximizing loan profit.)
Since no method for completely automating the solution to these problems are known, knowledge discovery is an interactive process, consisting of database interfaces, visualization of data, and automated pattern discovery (also known as data-mining). KDD thus combines methods from databases, computer graphics, artificial intelligence, and decision theory.