Discovery of constraints from data for information system reverse engineering

The extraction of functional dependencies is a fundamental activity in the database design recovery process which is part of an overall information systems reverse engineering effort. Existing algorithms for this task are computationally expensive and appear to be infeasible if applied to large legacy database instances, e.g., their performance deteriorated when number of attributes or/and instances is large and they cannot tolerate erroneous data that may occur in deployed commercial systems. The contributions of this paper are as follows. We propose two algorithms for discovering functional dependencies from data. The collective-FD algorithm, which is based on a top-down approach, eliminates redundant specialised functional dependencies to be proposed. The attribute-list algorithm, which is based on the bottom-up approach, enables more accurate functional dependency hypotheses to be discovered. In anticipating noisy data, we propose an effective method to discover possible data errors and compute partial functional dependencies. The result is an error-tolerant functional dependency discovery approach that is more applicable to real world databases for design recovery.