In recent years, there has been an emerging research area, called data mining or knowledge discovery, that addresses the problems in finding implicit, previously unknown, and potentially useful patterns in large databases. In this paper, we propose a top-down, interactive, and incremental method to discover previously unknown knowledge in large object-oriented databases which can be used in facilitating semantic query processing. Our approach consists of three phases: identification, extraction, and description. In the identification phase, we categorize or cluster objects into target classes which generalize or specialize groups of objects by identifying commonalities or differences among objects in a database. In the extraction phase, we discover potentially useful knowledge from each target class. In the description phase, we transform the discovered knowledge into the form of logic rules based on a first-order logic. The contribution of this paper is that we apply and extend data mining methods developed for relational databases to object-oriented databases with a deductive approach. In addition, we suggest methods to bias the search for interesting knowledge and to narrow the focus of the discovery process, which greatly decrease the number of objects considered for the discovery task and substantially reduces the computational complexity of the knowledge discovery process.
|Original language||English (US)|
|Number of pages||7|
|State||Published - Jan 1 1997|
ASJC Scopus subject areas
- Computer Science(all)