KMi Research Fellow Dr. Marco Ramoni has announced the launch of the latest version of his popular Bayesian Knowledge Discoverer (BKD). BKD is a data mining tool which can extract reusable knowledge from databases, using sound and accountable statistical methods, even when data is missing or incomplete. BKD Version 0.1 and Version 0.2 (Unix, AIX, and Mac versions) have already been distributed to over 1000 sites worldwide.

The capabilities of BKD Version 1.0 (Beta) include: estimation of conditional probability from data, extraction of the graphical structure from data, goal oriented evidence propagation, missing data handling using Bound and Collapse discretization of continous variables, automated definition of nodes from data, conversion from and to the proposed standard Bayesian Networks Interchance Format (BNIF), Graphic User Interface and movie-based on-line help.

System requirements

BKD Version 1.0 (Beta) for MS Windows 95/NT requires MS Windows 95/NT, 32 MB RAM (64 MB preferred), and 30 MB of free diskspace.

Details and download

Project details and a copy of BKD Version 1.0 (Beta) for MS Windows 95/NT can be obtained from the WWW site of the Bayesian Knowledge Discovery Project at http://kmi.open.ac.uk/projects/bkd