Anthony J. T. Lee

National Taiwan University, T’ai-pei, Taipei, Taiwan

Are you Anthony J. T. Lee?

Claim your profile

Publications (28)14.13 Total impact

  • Anthony J.T. Lee, Fu-Chen Yang, Hsin-Chieh Tsai, Yi-Yu Lai
    [Show abstract] [Hide abstract]
    ABSTRACT: Role analysis helps us characterize users’ interactions in a social network. However, previously proposed methods are mainly based on structural analysis of social networks rather than content-based behavior analysis. Therefore, we propose a method to use the content-based behavioral features extracted from user-generated content and behavior patterns to identify users’ roles and to explore role change patterns in social networks. The proposed method allows a user to play multiple roles in a social network and can identify roles without using any pre-defined roles. Thus, it provides a more general and flexible way to perform role analyses in social networks. The experimental results show that the proposed method can find various roles in different social networks, additional roles that haven’t been previously aware of, and some interesting role change patterns. The results may help us better understand the characteristics and trends of a social network, and formulate more effective management strategies.
    Decision Support Systems 01/2013; · 2.20 Impact Factor
  • Ying-Ho Liu, Anthony J.T. Lee, Fu Chang
    [Show abstract] [Hide abstract]
    ABSTRACT: The existing object recognition methods can be classified into two categories: interest-point-based and discriminative-part-based. The interest-point-based methods do not perform well if the interest points cannot be selected very carefully. The performance of the discriminative-part-base methods is not stable if viewpoints change, because they select discriminative parts from the interest points. In addition, the discriminative-part-based methods often do not provide an incremental learning ability. To address these problems, we propose a novel method that consists of three phases. First, we use some sliding windows that are different in scale to retrieve a number of local parts from each model object and extract a feature vector for each local part retrieved. Next, we construct prototypes for the model objects by using the feature vectors obtained in the first phase. Each prototype represents a discriminative part of a model object. Then, we establish the correspondence between the local parts of a test object and those of the model objects. Finally, we compute the similarity between the test object and each model object, based on the correspondence established. The test object is recognized as the model object that has the highest similarity with the test object. The experimental results show that our proposed method outperforms or is comparable with the compared methods in terms of recognition rates on the COIL-100 dataset, Oxford buildings dataset and ETH-80 dataset, and recognizes all query images of the ZuBuD dataset. It is robust enough for distortion, occlusion, rotation, viewpoint and illumination change. In addition, we accelerate the recognition process using the C4.5 decision tree technique, and the proposed method has the ability to build prototypes incrementally.
    Computer Vision and Image Understanding 07/2012; 116(7):854–867. · 1.23 Impact Factor
  • Anthony J T Lee, Ming-Chih Lin, Chia-Ming Hsu
    [Show abstract] [Hide abstract]
    ABSTRACT: Many methods have been proposed for mining protein complexes from a protein-protein interaction network; however, most of them focus on unweighted networks and cannot find overlapping protein complexes. Since one protein may serve different roles within different functional groups, mining overlapping protein complexes in a weighted protein-protein interaction network has attracted more and more attention recently. In this paper, we propose an effective method, called MDOS (Mining Dense Overlapping Subgraphs), for mining dense overlapping protein complexes (subgraphs) in a weighted protein-protein interaction network. The proposed method can integrate the information about known complexes into a weighted protein-protein interaction network to improve the mining results. The experiment results show that our method mines more known complexes and has higher sensitivity and accuracy than the CODENSE and MCL methods.
    Bio Systems 03/2011; 103(3):392-9. · 1.27 Impact Factor
  • Ming-Chih Lin, Anthony J. T. Lee, Rung-Tai Kao, Kuo-Tay Chen
    ACM Trans. Management Inf. Syst. 01/2011; 2:19.
  • Yao-Te Wang, Anthony J. T. Lee
    [Show abstract] [Hide abstract]
    ABSTRACT: Understanding the navigational behaviour of website visitors is a significant factor of success in the emerging business models of electronic commerce and even mobile commerce. However, Web traversal patterns obtained by traditional Web usage mining approaches are ineffective for the content management of websites. They do not provide the big picture of the intentions of the visitors. The Web navigation patterns, termed throughout-surfing patterns (TSPs) as defined in this paper, are a superset of Web traversal patterns that effectively display the trends toward the next visited Web pages in a browsing session. TSPs are more expressive for understanding the purposes of website visitors. In this paper, we first introduce the concept of throughout-surfing patterns and then present an efficient method for mining the patterns. We propose a compact graph structure, termed a path traversal graph, to record information about the navigation paths of website visitors. The graph contains the frequent surfing paths that are required for mining TSPs. In addition, we devised a graph traverse algorithm based on the proposed graph structure to discover the TSPs. The experimental results show the proposed mining method is highly efficient to discover TSPs.
    Expert Syst. Appl. 01/2011; 38:7112-7122.
  • Huei-Wen Wu, Anthony J. T. Lee
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we propose an efficient algorithm, called CFP, for mining closed flexible patterns in time-series databases, where flexible gaps are allowed in a pattern. Our proposed algorithm involves three stages: transforming a time-series database into a symbolic database, generating all frequent patterns of length one from the transformed database, and mining closed flexible patterns in a depth-first search manner. In the proposed method, we design two pruning strategies and a closure checking scheme to reduce the search space and thus speed up the algorithm. The experimental results show that our algorithm outperforms the modified Apriori algorithm by an order of magnitude.
    Expert Syst. Appl. 01/2010; 37:2098-2107.
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we proposed an efficient algorithm, called PCP-Miner (Pointset Closed Pattern Miner), for mining frequent closed patterns from a pointset database, where a pointset contains a set of points. Our proposed algorithm consists of two phases. First, we find all frequent patterns of length two in the database. Second, for each pattern found in the first phase, we recursively generate frequent closed patterns by a frequent pattern tree in a depth-first search manner. Since the PCP-Miner does not generate unnecessary candidates, it is more efficient and scalable than the modified Apriori, SASMiner and MaxGeo. The experimental results show that the PCP-Miner algorithm outperforms the comparing algorithms by more than one order of magnitude.
    Inf. Syst. 01/2010; 35:335-351.
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we propose a novel face detection method based on the MAFIA algorithm. Our proposed method consists of two phases, namely, training and detection. In the training phase, we first apply Sobel's edge detection operator, morphological operator, and thresholding to each training image, and transform it into an edge image. Next, we use the MAFIA algorithm to mine the maximal frequent patterns from those edge images and obtain the positive feature pattern. Similarly, we can obtain the negative feature pattern from the complements of edge images. Based on the feature patterns mined, we construct a face detector to prune non-face candidates. In the detection phase, we apply a sliding window to the testing image in different scales. For each sliding window, if the slide window passes the face detector, it is considered as a human face. The proposed method can automatically find the feature patterns that capture most of facial features. By using the feature patterns to construct a face detector, the proposed method is robust to races, illumination, and facial expressions. The experimental results show that the proposed method has outstanding performance in the MIT-CMU dataset and comparable performance in the BioID dataset in terms of false positive and detection rate.
    Pattern Recognition 01/2010; · 2.63 Impact Factor
  • Source
    Anthony J. T. Lee, Ming-Chih Lin, Rung-Tai Kao, Kuo-Tay Chen
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we propose an effective clustering method, HRK (Hierarchical agglomerative and Recursive K-means clustering), to predict the short-term stock price movements after the release of financial reports. The proposed method consists of three phases. First, we convert each financial report into a feature vector and use the hierarchical agglomerative clustering method to divide the converted feature vectors into clusters. Second, for each cluster, we recursively apply the K-means clustering method to partition each cluster into sub-clusters so that most feature vectors in each sub- cluster belong to the same class. Then, for each sub-cluster, we choose its centroid as the representative feature vector. Finally, we employ the representative feature vectors to predict the stock price movements. The experimental results show the proposed method outperforms SVM in terms of accuracy and average profits.
    Pacific Asia Conference on Information Systems, PACIS 2010, Taipei, Taiwan, 9-12 July 2010; 01/2010
  • Source
    Anthony J. T. Lee, Yi-An Chen, Weng-Chong Ip
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we propose an efficient graph-based mining (GBM) algorithm for mining the frequent trajectory patterns in a spatial–temporal database. The proposed method comprises two phases. First, we scan the database once to generate a mapping graph and trajectory information lists (TI-lists). Then, we traverse the mapping graph in a depth-first search manner to mine all frequent trajectory patterns in the database. By using the mapping graph and TI-lists, the GBM algorithm can localize support counting and pattern extension in a small number of TI-lists. Moreover, it utilizes the adjacency property to reduce the search space. Therefore, our proposed method can efficiently mine the frequent trajectory patterns in the database. The experimental results show that it outperforms the Apriori-based and PrefixSpan-based methods by more than one order of magnitude.
    Inf. Sci. 01/2009; 179:2218-2231.
  • Source
    Chun-sheng Wang, Anthony J. T. Lee
    [Show abstract] [Hide abstract]
    ABSTRACT: Sequential pattern and inter-transaction pattern mining have long been important issues in data mining research. The former finds sequential patterns without considering the relationships between transactions in databases, while the latter finds inter-transaction patterns without considering the ordered relationships of items within each transaction. However, if we want to find patterns that cross transactions in a sequence database, called inter-sequence patterns, neither of the above models can perform the task. In this paper, we propose a new data mining model for mining frequent inter-sequence patterns. We design two algorithms, M-Apriori and EISP-Miner, to find such patterns. The former is an Apriori-like algorithm that can mine inter-sequence patterns, but it is not efficient. The latter, a new method that we propose, employs several mechanisms for mining inter-sequence patterns efficiently. Experiments show that EISP-Miner is very efficient and outperforms M-Apriori by several orders of magnitude.
    Expert Syst. Appl. 01/2009; 36:8649-8658.
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we propose a novel algorithm, called 9DSPA-Miner, to mine frequent patterns from an image database, where every image is represented by the 9D-SPA representation. Our proposed method consists of three phases. First, we scan the database once and create an index structure. Next, the index structure is scanned to find all frequent patterns of length two. Finally, we use the frequent k-patterns (k⩾2) to generate candidate (k+1)-patterns and check if the support of each candidate generated is not less than the user-specified minimum support threshold by using the index structure. Then, the steps in the third phase are repeated until no more frequent patterns can be found. Since the 9DSPA-Miner algorithm uses the characteristics of the 9D-SPA representation to prune most of impossible candidates, the experiment results demonstrate that it is more efficient and scalable than the modified Apriori method.
    Journal of Systems and Software. 01/2009; 82:603-618.
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we propose an efficient algorithm, called CMP-Miner, to mine closed patterns in a time-series database where each record in the database, also called a transaction, contains multiple time-series sequences. Our proposed algorithm consists of three phases. First, we transform each time-series sequence in a transaction into a symbolic sequence. Second, we scan the transformed database to find frequent patterns of length one. Third, for each frequent pattern found in the second phase, we recursively enumerate frequent patterns by a frequent pattern tree in a depth-first search manner. During the process of enumeration, we apply several efficient pruning strategies to remove frequent but non-closed patterns. Thus, the CMP-Miner algorithm can efficiently mine the closed patterns from a time-series database. The experimental results show that our proposed algorithm outperforms the modified Apriori and BIDE algorithms.
    Data & Knowledge Engineering. 01/2009;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we propose an efficient algorithm, called ICMiner (Inter-transaction Closed patterns Miner), for mining closed inter-transaction itemsets. Our proposed algorithm consists of two phases. First, we scan the database once to find the frequent items. For each frequent item found, the ICMiner converts the original transaction database into a set of domain attributes, called a dataset. Then, it enumerates closed inter-transaction itemsets using an itemset–dataset tree, called an ID-tree. By using the ID-tree and datasets to mine closed inter-transaction itemsets, the ICMiner can embed effective pruning strategies to avoid costly candidate generation and repeated support counting. The experiment results show that the proposed algorithm outperforms the EH-Apriori, FITI, ClosedPROWL, and ITP-Miner algorithms in most cases.
    Data Knowl. Eng. 01/2008; 66:68-91.
  • Ping Yu, Anthony J. T. Lee, Naiwen Kuo, Chein-Shung Hwang
    [Show abstract] [Hide abstract]
    ABSTRACT: Modeling object's semantic knowledge has attracted increasing attention in the area of video content management. In this paper, we propose a video semantic model to get the higher-level semantics of spatial relation changes between the objects in a video represented by a 3D C-string, which represents the lower-level information of spatio-temporal relations, motions and size changes of the objects in a video. We use the concept of finite automata to record the transitions of objects' spatial relations. From the final states of the finite automaton, the higher-level semantics of spatial relation changes between the objects in a video can be inferred. Keywords: Video semantic model, Video content management, Spatial relation, 3D C-string
    Innovative Computing ,Information and Control, International Conference on. 01/2008;
  • Source
    Anthony J. T. Lee, Ping Yu, Han-Pang Chiu, Hsiu-Hui Lin
    J. Inf. Sci. Eng. 01/2007; 23:1-19.
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we propose a new filtration method, called Transformation-based Database Filtration method (TDF), to screen out those data sequences of a DNA sequence database which cannot satisfy a given query sequence. Our proposed method consists of two phases. First, we divide each data sequence into several windows (blocks), each of which is transformed into a data feature vector using the Haar wavelet transform. The transformed data feature vectors are then stored in an index file. Second, we divide a query sequence into sliding windows, each of which is, again, transformed into a query feature vector using the Haar wavelet transform. We then search the index file to find the candidate sequences for each query feature vector and check if they match the query sequence using the sequence alignment algorithm. We transform the bound of edit distance between sequences to the bound of Manhattan distance between feature vectors. Since the Manhattan distance is much easier to compute, our proposed method can efficiently screen out impossible data sequences and guarantee no false negatives. The experimental results show that our proposed method outperforms the QUASAR method in terms of filtration ratio, precision, execution time and index size. The proposed method also outperforms the YM method for long query, low complexity and repetitive data.
    Pattern Recognition Letters 01/2007; · 1.27 Impact Factor
  • Source
    Inf. Sci. 01/2007; 177:1593-1608.
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we propose a novel spatial mining algorithm, called 9DLT-Miner, to mine the spatial association rules from an image database, where every image is represented by the 9DLT representation. The proposed method consists of two phases. First, we find all frequent patterns of length one. Next, we use frequent k-patterns (k ⩾ 1) to generate all candidate (k + 1)-patterns. For each candidate pattern generated, we scan the database to count the pattern’s support and check if it is frequent. The steps in the second phase are repeated until no more frequent patterns can be found. Since our proposed algorithm prunes most of impossible candidates, it is more efficient than the Apriori algorithm. The experiment results show that 9DLT-Miner runs 2–5 times faster than the Apriori algorithm.
    Information Sciences. 01/2007;
  • Source
    Anthony J.T. Lee, Chun-Sheng Wang
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we propose an efficient method for mining all frequent inter-transaction patterns. The method consists of two phases. First, we devise two data structures: a dat-list, which stores the item information used to find frequent inter-transaction patterns; and an ITP-tree, which stores the discovered frequent inter-transaction patterns. In the second phase, we apply an algorithm, called ITP-Miner (Inter-Transaction Patterns Miner), to mine all frequent inter-transaction patterns. By using the ITP-tree, the algorithm requires only one database scan and can localize joining, pruning, and support counting to a small number of dat-lists. The experiment results show that the ITP-Miner algorithm outperforms the FITI (First Intra Then Inter) algorithm by one order of magnitude.
    Information Sciences. 01/2007;