TY - JOUR
T1 - Neutrosophic association rule mining algorithm for big data analysis
AU - Abdel-Basset, Mohamed
AU - Mohamed, Mai
AU - Smarandache, Florentin
AU - Chang, Victor
N1 - Publisher Copyright:
© 2018 by the authors.
PY - 2018/4/11
Y1 - 2018/4/11
N2 - Big Data is a large-sized and complex dataset, which cannot be managed using traditional data processing tools. Mining process of big data is the ability to extract valuable information from these large datasets. Association rule mining is a type of data mining process, which is indented to determine interesting associations between items and to establish a set of association rules whose support is greater than a specific threshold. The classical association rules can only be extracted from binary data where an item exists in a transaction, but it fails to deal effectively with quantitative attributes, through decreasing the quality of generated association rules due to sharp boundary problems. In order to overcome the drawbacks of classical association rule mining, we propose in this research a new neutrosophic association rule algorithm. The algorithm uses a new approach for generating association rules by dealing with membership, indeterminacy, and non-membership functions of items, conducting to an efficient decision-making system by considering all vague association rules. To prove the validity of the method, we compare the fuzzy mining and the neutrosophic mining. The results show that the proposed approach increases the number of generated association rules.
AB - Big Data is a large-sized and complex dataset, which cannot be managed using traditional data processing tools. Mining process of big data is the ability to extract valuable information from these large datasets. Association rule mining is a type of data mining process, which is indented to determine interesting associations between items and to establish a set of association rules whose support is greater than a specific threshold. The classical association rules can only be extracted from binary data where an item exists in a transaction, but it fails to deal effectively with quantitative attributes, through decreasing the quality of generated association rules due to sharp boundary problems. In order to overcome the drawbacks of classical association rule mining, we propose in this research a new neutrosophic association rule algorithm. The algorithm uses a new approach for generating association rules by dealing with membership, indeterminacy, and non-membership functions of items, conducting to an efficient decision-making system by considering all vague association rules. To prove the validity of the method, we compare the fuzzy mining and the neutrosophic mining. The results show that the proposed approach increases the number of generated association rules.
KW - Big data
KW - Data mining
KW - Neutrosophic association rule
KW - Neutrosophic sets
UR - http://www.scopus.com/inward/record.url?scp=85046087638&partnerID=8YFLogxK
U2 - 10.3390/sym10040106
DO - 10.3390/sym10040106
M3 - Article
AN - SCOPUS:85046087638
SN - 2073-8994
VL - 10
SP - 1
EP - 19
JO - Symmetry
JF - Symmetry
IS - 4
M1 - 106
ER -