Conclusion and future workin rough set

220 Zhongzhi Shi, Shaohui Liu, Zheng Zheng

In order to test the validity and ability of algorithm 4, we use the data of table 1 and also collect some data sets from UCI database, then we program based on our algorithm. Our experiment is carried on the computer whose frequency is PIII 800, EMS memory is 128M, and operating system is WIN2000. We compare the algorithms in reference [12] and [16] with our algorithm. is the algorithm in [12], is the algorithm in reference [16] and is ours (algorithm 4). We suppose n is the number of the objects, m is the number of condition attributes before reduced, is the number of condition attributes after reduced using and is the executing time of

In rough set, fuzzy problems are discussed by upper and lower approximation and we can use some relevant mathematical formula to get results. Rough set is objective and based on original data set totally. Nowadays, because of rough set’s unique feature, more and more scholars devoted into the research on it and made a great success. However, a lot of problems are needed to be resolved. Now, there are so little applications of rough set theory that have obvious benefits in industry[5], so it is important to find faster and more efficient algorithms to advance it.

In this paper, the properties of indiscernible relation are discussed and a new algorithm computing positive region is presented and proved. Based on these, we analyze the incrementally computing of positive region. Then, we develop an attribute reduction algorithm, which is complete and efficient with time complexity Our simulation results show that our algorithm is better than many existed attribute reduction algorithms. In addition, due to the efficiency of our algorithm, it is useful to the application of rough set and helpful to apply it in dealing with huge data set. The future work of this paper includes feature selecting with rough set, constructing a better classifier that is suitable to decision support and using other uncertain methods to improve the classifying ability.

1. Z. Pawlak. Rough Sets, International Journal of Computer and Information Science, 1982, 11(5): 341~356
2. Z. Pawlak. Rough Sets: Theoretical Aspects of Reasoning about Data, Dordrecht: Kluwer Academic Publishers, 1991
3. A. Skowron. Rough Sets and Boolean Reasoning, In: W. Pedrycz ed. Granular Computing: An emerging Paradigm, New York : Physica-Verlag, 2001
4. Q. Liu, Rough Set and Rough Reasoning, Scientific Press, 2001
5. W. Ziako. Rough Sets: Trends, Challenges, and Prospects, In: W. Ziarko, Y. Y. Yao ed.

Rough Sets and Current Trends in Computing(RSCTC 2000), Berlin: Springer-verlag, 2001
6. Q.Liu, S.H. Huang, L.W Yao, Rough Set Theory: Present State And Prospects, Computer Science, 1997, 24(4): 1~5
7. J. Wang, D.Q.Miao, Rough Set Thoery and Its Application: A Survey, Pattern Recognition And Artificial Intelligence, 1996, 9(4): 337~344
8. A. Skowron, C. Rauszer. The Discernibility Matrics and Functions in Inforamtion System.