Construct Bayesian Networks from Functional Dependencies

H. Bian and S. Liao (PRC)


database tools and software, data mining, Bayesian network, functional dependency


Functional dependency is an important data dependency and information resource in conventional relational database. Bayesian network is a popular technique in probabilistic representing and reasoning. In this paper we present BN-FD, a system prototype that constructs Bayesian networks, which can be used as data mining tools, from functional dependencies defined in database schemas or from those found in data. Compared with the traditional statistical-based approach for constructing Bayesian networks from data, our approach is more efficient and capable of using existing information in conventional databases. Building a Bayesian network includes constructing its structure and setting prior probabilities. At this stage, this paper is focused on the construction of the structure of a Bayesian network from the functional dependencies obtained.

Important Links:

Go Back