Interesting Rough Lattice-based Implication Rules Discovery
Yi Zhao, Jianliang Shi, Da Ruan, Pengfei Shi · Studies in fuzziness and soft computing · 2001
As an important data mining technique, implication rules help to explore the dependencies among attributes of a database. In this paper, we aim at finding an efficient algorithm to discover the implication rules in a data set. This algorithm first extends the concept lattice theory by building the lattice according to the data set with the interesting attributes to improve user interaction and mining efficiency. The interesting concept lattice, together with the rough set theory, is then incorporated into our method to implement a new interesting rough lattice-based implication rules discovery (IRLIRD) approach to interactively acquire the rules with the specific rough upper and lower approximation. It generates implication rules without multiple passes over the data set and computing frequent itemsets. For the application of the proposed method to the transaction data set of the large-scale supermarkets, a novel data structure, which is based on the numeric attribute and termed as linked list of transaction set, is introduced here to reduce the needed memory. A simulation is implemented to illustrate the whole mining process, which demonstrates that the approach reduces the computational time greatly comparing with that of the classical Apriori algorithm. The algorithm can also be extended to many other application areas such as stock analysis, credit card distribution and agricultural application, etc. These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.