An algorithm of multi-level fuzzy association rules mining with multiple minimum supports in network faults diagnosis

In business applications, there have been tremendous interests in analysing customers' repeated purchase behaviour. Recently, the concepts of periodic pattern and cyclic pattern are used to discover recurring patterns from customer sequence database. Toroslu (2003) proposed cyclic pattern mining, which considers a new parameter, named repetition support, into the mining process. In a customer sequence, the occurrence of a subsequence must satisfy single user-specified repetition minimum support. In real-life applications, however, different items may have different frequencies in the database. If all items are set to have the same minimum repetition support, it may cause rare item problem. To solve this problem, we include the concept of multiple minimum supports (MMS) to allow users to specify multiple minimum item repetition support (MIR) according to the natures of items. In this paper, we first redefine cyclic sequential patterns based on MIR and original form of customer minimum support. A new algorithm, rep-PrefixSpan, is developed to discover complete set of cyclic sequential patterns from sequence database. The experimental result shows that the proposed approach achieves more preferable findings than conventional cyclic pattern mining.

Association rules mining is an important research topic in data mining and knowledge discovery. Traditional algorithms for mining association rules are built on the binary attributes databases, which has three limitations. Firstly, it can not concern quantitative attributes; secondly, only the positive association rules are discovered; thirdly, it treat each item with the same frequency although different item may have different frequency. In this paper, we put forward a discovery algorithm for mining positive and negative fuzzy association rules to resolve these three limitations.

In real-world applications, transactions are typically represented by quantitative data. Thus, fuzzy association rule mining algorithms have been proposed to handle these quantitative transactions. In addition, items generally have certain lifespans or temporal periods in which they exist in a database. Therefore, fuzzy temporal association rule mining algorithms have also been proposed in the literature. A key factor in the acquisition of fuzzy temporal association rules (FTARs) is the design of appropriate membership functions. Because current approaches have been designed to generate membership functions for mining fuzzy association rules (FARs) in market-basket analysis, in this paper, we propose a membership function tuning mechanism for a fuzzy temporal association rule mining algorithm. The proposed approach modifies an existing cluster-based method to generate unique membership functions that are specifically tailored to each item in a dataset. Two factors are utilized to decide the appropriate membership functions of each item: (1) the density similarity among intervals corresponding to the density similarity within intervals, and (2) the information closeness within an interval corresponding to the similarity in the number of data points between intervals. A parameter θ is used to indicate the relative importance of these two factors. As a result, the membership functions are generated based on the quantitative ranges of individual items, and the generated membership functions of items are different in terms of the values of each interval and the number of intervals. The generated membership functions are subsequently used in a fuzzy temporal association rule mining algorithm. Computational experiments were conducted on both a synthetic dataset and a real-world one to demonstrate the effectiveness of the proposed approach.

Fault diagnosis is a challenging work in large communication network. A fault will bring more than one alarms, an alarm may be caused by different faults vice versa. The relationship between faults and alarms are not accurate but fuzzy, which can not be understood using traditional Boolean logic. Also, crisp association rules use sharp partitioning and potentially introduce loss of information. Therefore, fuzzy logic is used to convert numerical attributes into linguistic terms and a fuzzy judge language set is described. We first analyze the meanings of each attribute of network alarms. Then define proper fuzzy membership functions according to the relationship between this field and the root alarm. After that we integrate MDs (membership degrees) and weights of every attribute to construct fuzzy database. At last, we come up with a new fuzzy association rules mining algorithm, which generalizes the popular hierarchy-based algorithm. Simulation experiments on a synthetic network are carried out to prove and validate the accuracy and efficiency of our algorithm with respect to correlation analysis for right positioning root alarms.

Traditional algorithms for mining sequential patterns are built on the binary attributes databases, which have three limitations. Firstly, it can not concern quantitative attributes; secondly, only direct sequential patterns are discovered; thirdly, it can not process these data items with similar frequencies which will result in the dilemma called the rare item problem. In this paper, we put forward a discovery algorithm for mining both direct and indirect fuzzy sequential patterns with multiple minimum supports by combining these three extensions.

Mining generalized association rules among items in the presence of taxonomy and with nonuniform minimum support has been recognized as an important model in the data mining community. In real applications, however, the work of discovering interesting association rules is an iterative process; the analysts have to continuously adjust the constraint of minimum support to discover real informative rules. How to reduce the response time for each remining process thus becomes a crucial issue. In this paper, we examine the problem of maintaining the discovered multi-supported generalized association rules when the multiple minimum support constraint is refined and propose a novel algorithm called RGA_MSR to accomplish the work. By keeping and utilizing the set of frequent itemsets and negative border, and adopting vertical intersection counting strategy, the proposed RGA_MSR algorithm can significantly reduce the computation time spent on rediscovery of frequent itemsets and has very good performance.

A major problem when conducting mining fuzzy association rules from the database (DB) is the large computation time and memory needed. In addition, the selection of fuzzy sets for each attribute of the database is very important because it will affect the quality of the mining rule. This paper proposes a method for mining fuzzy association rules using the compressed database. We also use the approach of Hedge Algebra (HA) to build the membership function for attributes instead of using the normal way of fuzzy set theory. This approach allows us to explore fuzzy association rules through a relatively simple algorithm which is faster in terms of time, but it still brings association rules which are as good as the classical algorithms for mining association rules.

Association rule is an important research topic in data mining and knowledge discovery. Traditional algorithms for mining association rules are built on the binary attributes databases, which has three limitations. Firstly, it can not concern quantitative attributes; secondly, it treats each item with the same significance although different item may have different significance; thirdly, only the direct association rules are discovered. Mining fuzzy association rules has been proposed to address the first limitation. In this paper, we put forward an idea for mining indirect weighted association rules to resolve the other two limitations, and a discovery algorithm for mining both direct and indirect weighted fuzzy association rules by integrating these three extensions.

We have been using fuzzy data mining techniques to extract patterns that represent behavior for intrusion detection. We describe a variety of modifications that we have made to the data mining algorithms in order to improve accuracy and efficiency. We use sets of fuzzy association rules that are mined from network audit data as models of behavior. To detect anomalous behavior, we generate fuzzy association rules from new audit data and compute the similarity with sets mined from normal data. If the similarity values are below a threshold value, an alarm is issued. We describe an algorithm for computing fuzzy association rules based on Borgelt's (2001) prefix trees, modifications to the computation of support and confidence of fuzzy rules, a new method for computing the similarity of two fuzzy rule sets, and feature selection and optimization with genetic algorithms. Experimental results demonstrate that we can achieve better running time and accuracy with these modifications.

In this paper, an enhanced efficient approach for speeding up the evolution process for finding minimum supports, membership functions and fuzzy association rules is proposed by utilizing clustering techniques. All the chromosomes use the requirement satisfaction derived only from the representative chromosomes in the clusters and from their own suitability of membership functions to calculate the fitness values. The evaluation cost can thus be greatly reduced due to the cluster-based time-saving process. The final best minimum supports and membership functions in all the populations are then gathered together for mining fuzzy association rules. Experimental results also show the efficiency of the proposed approach.

Association rule mining ARM algorithms work only with binary attributes, and expect quantitative attributes to be converted to binary ones using sharp partitions, like 'age = [25, 60]'. A better alternative is to convert quantitative attributes to fuzzy attributes, like 'age = middle-aged', to eliminate loss of information due to sharp partitioning, and then run a fuzzy ARM algorithm. The most popular fuzzy ARM algorithms are fuzzy adaptations of apriori. Fuzzy apriori, like apriori, is a slow algorithm, especially for most medium-sized 500 K to 1 M and large > 1 M datasets. We propose a new fuzzy ARM algorithm called FAR-miner for fast and efficient performance. Through experiments we show that FAR-miner is 8-19 and 6-10 times faster on large and medium-sized datasets respectively as compared to fuzzy apriori. This efficiency is due to properties like two-phased multiple-partition tidlist-style processing and byte-vector representation and effective compression of tidlists.

Mining association rules with multiple minimum supports: a new mining algorithm and a support tuning mechanism

Fuzzy ellipsoid theory was introduced to mining fuzzy association rule from quantitative data. The algorithm includes two steps. First, fuzzy clustering procedure is taken to get fuzzy ellipsoids of the feature space. Second, mining procedure is taken to get fuzzy association rule based on the projection regions of fuzzy ellipsoids. The algorithm was compared with a Apriori-based fuzzy association rule mining algorithm. Results show that the algorithm is considerably quicker, especially when mining from large databases. Experiments on real-life data also show that the algorithm is effective to find association rules.

Data Mining denotes mining knowledge from hugequantity of data. All algorithms of association rules mining include ‘first finding frequency of item sets, which accept a minimum support threshold, and then calculates confidence percentage for all k-item sets to construct robust association rules’. The trouble is there are some of algorithms that need more time for compute minimum support, minimum confidence and extraction larger item. In this paper one algorithm is proposed (enhanced reduces items Apriori algorithm) to reduce execution time. The proposed algorithm purpose to introduce algorithm to mine association rules to obtain fast algorithm by reducing execute time. Due to many experiments in (enhanced reduces items Apriori algorithm), this algorithm is very fast compared with (topk-rules and topk-non redundant rules) algorithms.

In traditional association rule mining algorithms, if the minimum support is set too high, many valuable rules will be lost. However, if the value is set too low, then numerous trivial rules will be generated. To overcome the difficulty of setting minimum support values, global and local patterns are mined herein. Owing to the temporal factor in association rule mining, an itemset may not occur frequently in the entire dataset (meaning that it is not a global p attern), but it may appear frequently over specific intervals (meaning that it is a local pattern). This paper proposed a temporal ass ociation rule mining algorithm for interval frequent-patterns, called GLFMiner, which automatically and efficiently generates all intervals w ithout prior domain knowledge in an efficient manner. GLFMiner considers not only global frequent-patterns, but also local frequent-patterns. Using the same value of minimum support, it can locate many valuable temporal rules without losing the rules that traditio nal algorithms may find. Experimental results reveal that our novel algorithm mines more temporal frequent-patterns than traditional association rule mining algorithms and is effective in real-world applications such as market basket analysis and intrusion detection systems.