Ja-Hwung Su

A fast algorithm for mining high average-utility itemsets

Mining high-utility itemsets (HUIs) in transactional databases has become a very popular research topic in recent years. A popular variation of the problem of HUI mining is to discover high average-utility itemsets (HAUIs), where an alternative measure called the average-utility is used to evaluate the utility of itemsets by considering their lengths. Albeit, HAUI mining has been studied extensively, current algorithms often consume a large amount of memory and have long execution times, due to the large search space and the usage of loose upper bounds to estimate the average-utilities of itemsets. In this paper, we present a more efficient algorithm for HAUI mining, which includes three pruning strategies to provide a tighter upper bound on the average-utilities of itemsets, and thus reduce the search space more effectively to decrease the runtime. The first pruning strategy utilizes relationships between item pairs to reduce the search space for itemsets containing three or more items. The second pruning strategy provides a tighter upper bound on the average-utilities of itemsets to prune unpromising candidates early. The third strategy reduces the time for constructing the average-utility-list structures for itemsets, which is used to calculate their upper bounds. Substantial experiments conducted on both real-life and synthetic datasets show that the proposed algorithm with three pruning strategies can efficiently and effectively reduce the search space for mining HAUIs, when compared to the state-of-the-art algorithms, in terms of runtime, number of candidates, memory usage, performance of the pruning strategies and scalability.

Efficient Mining of High Average-Utility Itemsets with Multiple Minimum Thresholds

High average-utility itemsets mining (HAUIM) is a key data mining task, which aims at discovering high average-utility itemsets (HAUIs) by taking itemset length into account in transactional databases. Most of these algorithms only consider a single minimum utility threshold for identifying the HAUIs. In this paper, we address this issue by introducing the task of mining HAUIs with multiple minimum average-utility thresholds (HAUIM-MMAU), where the user may assign a distinct minimum average-utility threshold to each item or itemset. Two efficient IEUCP and PBCS strategies are designed to further reduce the search space of the enumeration tree, and thus speed up the discovery of HAUIs when considering multiple minimum average utility thresholds. Extensive experiments carried on both real-life and synthetic databases show that the proposed approaches can efficiently discover the complete set of HAUIs when considering multiple minimum average-utility thresholds.

Using grouping genetic algorithm to mine diverse group stock portfolio

In this paper, to increase the diversity of stock portfolios, the diverse group stock portfolio mining algorithm is proposed based on the grouping genetic algorithm. Each chromosome is represented by grouping part, stock part and stock portfolio part. The fitness function that consists of portfolio satisfaction, group balance and diversity factor is designed to evaluate quality of chromosomes. The diversity factor is used to make the numbers of stock categories in groups as similar as possible. The genetic operations are then executed on population to generate offspring for finding a near-optimal group stock portfolio. Finally, experiments on a real financial data were made to show the effectiveness of the proposed approach.

An Item-Based Music Recommender System Using Music Content Similarity

Nowadays, music data grows rapidly because of the advanced multimedia technology. People are always spending much time to listen to music. This incurs a hot research issue for how to discover the users’ interested music preferences from a large amount of music data. To deal with this issue, the music recommender system has been a solution that can infer the users’ musical interests by a set of learning methods. However, recent music recommender systems encounter problems of new item and data sparsity. To alleviate these problems, in this paper, we propose a new recommender system that fuses user ratings and music low-level features to enhance the recommendation quality. The experimental results show that our proposed recommender system outperforms other well-known music recommender systems.

Mining and applications of repeating patterns

Mining the valuable knowledge from real data has been a hot topic for a long time. Repeating pattern is one of the important knowledge, occurring in many real applications such as musical data and medical data. In this paper, our purposes are to contribute an efficient mining algorithm for repeating patterns and to conduct a real application using the repeating patterns mined. In terms of mining the repeating patterns, although a number of past studies were made on this issue, the performance cannot still earn the users’ satisfactions especially for large data sets. For this issue, in this paper, we propose an efficient algorithm named Fast Mining of Repeating Patterns, which achieves high performance of discovering the repeating patterns by a novel index called Quick-Pattern Index. In terms of applications, a music recommender system named repeating-pattern-based music recommender system is proposed to deal with problems in music recommendation. Even facing a very sparse rating matrix, the recommendation can still be completed. The experimental results show that our proposed mining algorithm and recommender system outperform the previous works in terms of efficiency and effectiveness, respectively.

An island-based algorithm for group stock portfolio optimization

This paper presents an island-based optimization approach to speed up the evaluation process for optimizing a diverse group stock portfolio which can provide various chooses for users to make investment decisions. It first initializes a population for each island. For every t generation, the best chromosome of each island is selected and putted into a master island. Then, chromosomes in master island will migrate to other islands in order to get a near optimal solution. Experimental results on a real dataset were also conducted and indicated that the proposed approach is better than the previous approach in terms of evolution time and returns of the optimized stock portfolios.

A High-Performance Algorithm for Mining Repeating Patterns

A repeating pattern is a sequence composed of identical elements, repeating in a regular manner. In real life, there are lots of applications such as musical and medical sequences containing valuable repeating patterns. Because the repeating patterns hidden in sequences might contain implicit knowledge, how to retrieve the repeating patterns effectively and efficiently has been a challenging issue in recent years. Although a number of past studies were proposed to deal with this issue, the performance cannot still earn users’ satisfactions especially for large datasets. To aim at this issue, in this paper, we propose an efficient algorithm named Fast Mining of Repeating Patterns (FMRP), which achieves high performance for finding repeating patterns by a novel index called Quick-Pattern-Index (QPI). This index can provide the proposed FMRP algorithm with an effective support due to its information of pattern positions. Without scanning a given sequence iteratively, the repeating patterns can be discovered by only one scan of the sequence. The experimental results reveal that our proposed algorithm performs better than the compared methods in terms of execution time.

More Efficient Algorithm to Mine High Average-Utility Patterns

In this paper, an efficient algorithm with three pruning strategies are presented to provide tighter upper-bound average-utility of the itemsets, thus reducing the search space for mining the set of high average-utility itemsets (HAUIs). The first strategy finds the relationships of the 2-itemsets, thus reducing the search space of k-itemsets (k ≥ 3). The second and the third pruning strategies set lower upper-bounds of the itemsets to early reduce the unpromising candidates. Substantial experiments show that the proposed algorithm can efficiently and effectively reduce the search space compared to the state-of-the-art algorithms in terms of runtime and number of candidates.

Efficient mining of short periodic high-utility itemsets

Mining of high-utility itemsets in transactional databases is emerging topic in recent years since it can be used to reveal more information for decision making, which has been widely used in many real-life applications. For the traditional high-utility itemset mining (HUIM), only the utility values of the itemsets are considered without timestamps or periodic constraints. In this paper, we present a new short periodic high-utility itemset mining (SPHUIM) to mine the set of short periodic high-utility itemsets (SPHUIs) by considering both the period and utility measures. A baseline two-phase SPHUI-TP algorithm is first presented to mine SPHUIs in level-wise manner. To reduce the search space of SPHUI-TP algorithm, two pruning strategies are also developed to speed up the mining performance of the SPHUI-TP algorithm. Substantial experiments both on real-life and synthetic datasets showed the efficiency and effectiveness of the designed approaches.

Fast algorithms for mining multiple fuzzy frequent itemsets

In the past, several algorithms were developed to mine fuzzy frequent itemsets (FFIs) in which each item is represented at most one linguistic term based on maximum scalar cardinality. In real-life situations, multiple fuzzy linguistic terms instead of the single one can, however, produce more useful and meaningful fuzzy association rules. The Apriori-based algorithm was developed to mine multiple fuzzy frequent itemsets (MFFIs), which requires to generate the amounts of candidates and determine them in a level-wise way. In this paper, a fuzzy-list-based (FL)-Miner algorithm is developed to mine the complete set of MFFIs without candidate generation. Two efficient pruning strategies are also developed to reduce the search space, thus speeding up the mining process to directly discover the MFFIs. Experiments are conducted to show the performance of the proposed approaches compared to the state-of-the-art level-wise algorithm in terms of execution time and memory usage.