Xiaowei Yan

Genetic algorithm-based strategy for identifying association rules without specifying actual minimum support

We design a genetic algorithm-based strategy for identifying association rules without specifying actual minimum support. In this approach, an elaborate encoding method is developed, and the relative confidence is used as the fitness function. With genetic algorithm, a global search can be performed and system automation is implemented, because our model does not require the user-specified threshold of minimum support. Furthermore, we expand this strategy to cover quantitative association rule discovery. For efficiency, we design a generalized FP-tree to implement this algorithm. We experimentally evaluate our approach, and demonstrate that our algorithms significantly reduce the computation costs and generate interesting association rules only.

ARMGA: IDENTIFYING INTERESTING ASSOCIATION RULES WITH GENETIC ALGORITHMS

ABSTRACT Priori-like algorithms for association rules mining have relied on two user-specified thresholds: minimum support and minimum confidence. There are two significant challenges to applying these algorithms to real-world applications: database-dependent minimum-support and exponential search space. Database-dependent minimum-support means that users must specify suitable thresholds for their mining tasks though they may have no knowledge concerning their databases. To circumvent these problems, in this paper, we design an evolutionary mining strategy, namely the ARMGA model, based on a genetic algorithm. Like general genetic algorithms, our ARMGA model is effective for global searching, especially when the search space is so large that it is hardly possible to use deterministic searching method.

Toward databases mining: pre-processing collected data

This paper presents a new means of selecting quality data for mining multiple data sources. Traditional data-mining strategies obtain necessary data from internal and external data sources and pool all the data into a huge homogeneous dataset for discovery. In contrast, our data-mining strategy identifies quality data from (internal and external) data sources for a mining task. A framework is advocated for generating quality data. Experimental results demonstrate that application of this new data collecting technique can not only identify quality data, but can also efficiently reduce the amount of data that must be considered during mining.

ON DATA STRUCTURES FOR ASSOCIATION RULE DISCOVERY

Systematically we study data structures used to implement the algorithms of association rule mining, including hash tree, itemset tree, and FP-tree (frequent pattern tree). Further, we present a generalized FP-tree in an applied context. This assists in better understanding existing association-rule-mining strategies. In addition, we discuss and analyze experimentally the generalized k-FP-tree, and demonstrate that the generalized FP-tree reduces the computation costs significantly. This study will be useful to many association analysis tasks where one must provide really interesting rules and develop efficient algorithms for identifying association rules.

CONFIDENCE METRICS FOR ASSOCIATION RULE MINING

We propose a simple, novel, and yet effective confidence metric for measuring the interestingness of association rules. Distinguishing from existing confidence measures, our metrics really indicate the positively companionate correlations between frequent itemsets. Furthermore, some desired properties are derived for examining the goodness of confidence measures in terms of probabilistic significance. We systematically analyze our metrics and traditional ones, and demonstrate that our new algorithm significantly captures the mainstream properties. Our approach will be useful to many association analysis tasks where one must provide actionable association rules and assist users to make quality decisions.

IDENTIFYING SOFTWARE COMPONENT ASSOCIATION WITH GENETIC ALGORITHM

Identifying software component association is useful for component management and component retrieval. In this paper we design an evolutionary strategy to understand software structure better and identify software component association, by using genetic algorithm. Our mining strategy is effective for global search, especially when the searched space is so large that it is hardly possible to use deterministic search method.

Identifying frequent terms in text databases by association semantics

Existing information retrieval methods are mainly based on either term similarity or latent semantics. To reduce irrelevant information searched, this paper presents a new approach for information retrieval by applying the methodology of association rule mining to a text database. Association semantics among terms of a document and a query are considered, such that the semantic similarity between the document and query may be reduced if they are somewhat irrelevant.