Mining multidimensional fuzzy association rules of alarms in communication networks

In the modern communication network, a fault will bring more than one alarm, an alarm may be caused by different faults vice versa. The relationship between faults and alarms is not accurate but fuzzy, which can not be described and understood using traditional Boolean logic. Also, crisp association rules use sharp partitioning to transfer numerical attributes to binary ones, and can potentially introduce loss of information due to these sharp ranges. As fuzzy sets provide a smooth transition between member and non-member of a set, fuzzy association rules use fuzzy logic to convert numerical attributes to linguistic terms. We first analyze the meanings of each field of network alarms. Then define a fuzzy judge language set as domain set to describe the relationship between the field and the root alarm. After that we integrate fuzzy membership function values and weights of every field of alarm to change alarm database to fuzzy alarm database. At last, we come up with a new fuzzy association rules mining algorithm, which generalizes the popular frequent itemsets based algorithm. The advantages and efficiency of the new algorithm are shown by experiments on a communication network database with alarm transaction records.