Discovery of First-Order Regularities in a Relational
Database Using Offline Candidate Determination

Irene Weber

Institut fr Informatik, Universitt Stuttgart,
Breitwiesenstr. 2022, 70565 Stuttgart, Germany
email: Irene.Weber@informatik.uni-stuttgart.de



Abstract. In this paper, we present an algorithm for the discovery of first order 
clauses holding in an relational database in the framework of the nonmonotonic 
ILP setting [1]. The algorithm adopts the principle of offline candidate
determination algorithm used for mining association rules in large transaction
databases [4]. Analoguous to the measures used in mining association rules, we
define a support and a confidence measure as acceptance criteria for discovered
hypothesis clauses.
The algorithm has been implemented in C with an interface to the relational
database management system INGRES. We present and discuss the results of
an experiment in the KRK domain and conclude.
References

1.	L. De Raedt and M. Bruynooghe. A theory of clausal discovery. In Proc. of IJCAI-9S,
Chambry, 1993.
2.	U. Fayyad, G. Piatetsky-Shapiro, P. Smyth, and R. Uthurusamy. Advances in Knowledge
Discovery and Data Mining. MIT Press, Cambridge, MA, 1995.
3.	G. Lindner and K. Morik. Coupling a relational learning algorithm with a database
system. In Y. Kodratoff, G. Nakhaeizadeh, and C. Taylor, editors, Workshop Notes
of the ECML-95 Workshop Statistics, Machine Learning and Knowledge Discovery in
Databases, 1995.
4.	H. Mannila, H. Toivonen, and I. Verkamo. Efficient algorithms for discovering association
rules. In AAAI-94 Workshop on Knowledge Discovery in Databases, 1994.
