2nd European Workshop on Data Mining and Text Mining for Bioinformatics, Piacenza, Italy, 24 - 26 September 2004, pp.35-41
Recent advancement in biotechnology has produced a massive amount of raw
biological data which are accumulating at an exponential rate. Errors,
redundancy and discrepancies are prevalent in the raw data, and there is
a serious need for systematic approaches towards biological data
cleaning. This work examines the extent of redundancy in biological data
and proposes a method for detecting duplicates in biological data.
Duplicate relations in a real-world biological dataset are modeled into
forms of association rules so that these duplicate relations or rules
can be induced from data with known duplicates using association rule
mining. Our approach of using association rule induction to find
duplicate relations is new. Evaluation of our method on a real-world
dataset shows that our duplicate association rules can accurately
identify up to 96.8% of the duplicates in the dataset at the accuracy of
0.3% false positives and 0.0038% false negatives.