Incorporating occupancy into frequent pattern mining for high quality pattern recommendation

Abstract

Mining interesting patterns from transaction databases has attracted a lot of research interest for more than a decade. Most of those studies use frequency, the number of times a pattern appears in a transaction database, as the key measure for pattern interestingness. In this paper, we introduce a new measure of pattern interestingness, occupancy. The measure of occupancy is motivated by some realworld pattern recommendation applications which require that any interesting pattern X should occupy a large portion of the transactions it appears in. Namely, for any supporting transaction t of pattern X, the number of items in X should be close to the total number of items in t. In these pattern recommendation applications, patterns with higher occupancy may lead to higher recall while patterns with higher frequency lead to higher precision. With the definition of occupancy we call a pattern dominant if its occupancy is above a user-specified threshold. Then, our task is to identify the qualified patterns which are both frequent and dominant. Additionally, we also formulate the problem of mining top-k qualified patterns: finding the qualified patterns with the top-k values of any function (e.g. weighted sum of both occupancy and support).The challenge to these tasks is that the monotone or anti-monotone property does not hold on occupancy. In other words, the value of occupancy does not increase or decrease monotonically when we add more items to a given itemset. Thus, we propose an algorithm called DOFIA (DOminant and Frequent Itemset mining Algorithm), which explores the upper bound properties on occupancy to reduce the search process. The tradeoff between bound tightness and computational complexity is also systematically addressed. Finally, we show the effectiveness of DOFIA in a real-world application on print-area recommendation for Web pages, and also demonstrate the efficiency of DOFIA on several large synthetic data sets.

An Evolutionary Algorithm for Large-Scale Sparse Multiobjective Optimization Problems

“…Due to the exponentially increased search space of LOPs, they cannot be easily solved by general evolutionary algorithms, hence some novel techniques have been adopted to tackle specific types of LOPs, such as variable interaction analysis [32]- [34], linkage learning [35], and random embedding based Bayesian optimization [36], [37]. For the LOPs in many important fields including machine learning [18], data mining [10], and network science [11], many problems contain sparse optimal solutions. However, no technique has been tailored for sparse problems before.…”

“…First, many sparse SOPs embed the sparsity of solutions in the objective function [15], [39], which can actually be regarded as the second objective to be optimized; that is, sparse SOPs can be converted into sparse MOPs. Second, the optimal solutions of some problems are sparse only if multiple conflicting objectives are considered [2], [10].…”

“…Four popular sparse MOPs in applications are introduced in the following, namely, feature selection [8], pattern mining [10], critical node detection [11], and neural network training [18]. These four problems are described in Fig.…”

“…Figs. 10 and 11 plot the results on PM4 and CN4, where the results obtained by a traditional pattern mining approach DOFIA [10] and a traditional critical node detection approach SEQ [80] are also shown in the figures. It turns out that the solution sets obtained by SparseEA have better convergence and diversity than those obtained by the traditional approaches and the compared MOEAs.…”

“…Therefore, if binary encoding is adopted (i.e., each binary decision variable indicates whether one feature is selected or not), most decision variables of the Pareto optimal solutions will be zero, in other words, the Pareto optimal solutions are sparse. Such subset selection problems with sparse Pareto optimal solutions are commonly seen in many other applications, for instance, sparse regression [9], pattern mining [10], and critical node detection [11].…”