Helen S. Du

Weblog success: Exploring the role of technology

Weblogs have recently gained considerable media attention. Leading weblog sites are already attracting millions of visitors. Yet, success in the highly competitive world of weblogs is not easily achieved. This study seeks to explore weblog success from a technology perspective, i.e. from the impact of weblog-building technology (or blogging tool). Based on an examination of 126 highly successful weblogs tracked over a period of 3 months, we categorized weblogs in terms of popularity rank and growth, and evaluated the relationship between weblog success (in terms of popularity) and technology use. Our analysis indicates that weblog success is associated with the type of blogging tool used. We argue that technology characteristics affect the presentation and organization of weblog content, as well as the social interaction between bloggers, and in turn, affect weblog success or popularity improvement. Based on this analysis, we propose a techno-social success model for weblogs. This model postulates that a weblogs success is mainly associated with its ability to provide value for its users and readers at the content, the technology, and the social levels.

Learning with Weblogs: An Empirical Investigation

The study investigates the impact of weblog use on individual learning in a university environment. Weblogs are a relatively new knowledge sharing technology, which enables people to record their thoughts in diary form and publish those diaries as web pages, without programming or HTML coding. The research sought to empirically determine whether the keeping of on-going (web based) learning logs throughout a semester would result in better overall student performance. This was hypothesized, because web based learning logs appear to promote constructivist learning, provide reinforcement, and increase accountability (non-anonymous idea sharing). Results from an information systems undergraduate course with 31 students indicate that weblog performance is a significant predictor for learning outcome, while traditional coursework is not. Weblogs appear to have highest predictive power for high and low performing students, but much less predictive value for medium performers. Results also suggest that there is a learning effect for weblog authoring.

Feature Extraction Using Laplacian Maximum Margin Criterion

Feature extraction by Maximum Margin Criterion (MMC) can more efficiently calculate the discriminant vectors than LDA, by avoiding calculation of the inverse within-class scatter matrix. But MMC ignores the local structures of samples. In this paper, we develop a novel criterion to address this issue, namely Laplacian Maximum Margin Criterion (Laplacian MMC). We define the total Laplacian matrix, within-class Laplacian matrix and between-class Laplacian matrix by using the similar weight of samples to capture the scatter information. Laplacian MMC based feature extraction gets the discriminant vectors by maximizing the difference between between-class laplacian matrix and within-class laplacian matrix. Experiments on FERET and AR face databases show that Laplacian MMC works well.

XML-BASED DECISION SUPPORT SYSTEMS: CASE STUDY FOR PORTFOLIO SELECTION

The extensible, structural and validated nature of XML provides standard data representation for efficient data interchange among diverse information resources available on the Web. Therefore, it leads to its growing recognition in e-commerce and Internet-based information exchange. In this paper, we stress the adoption of XML technology in developing efficient and flexible Web-enabled decision support systems. Based on a case study for portfolio selection systems, we explore the design issues in applying XML to overcome the heterogeneity of data exchange and sharing of various portfolio optimization models.

Face Recognition Using Kernel UDP

UDP has been successfully applied in many fields, finding a subspace that maximizes the ratio of the nonlocal scatter to the local scatter. But UDP can not represent the nonlinear space well because it is a linear method in nature. Kernel methods can otherwise discover the nonlinear structure of the images. To improve the performance of UDP, kernel UDP (a nonlinear vision of UDP) is proposed for face feature extraction and face recognition via kernel tricks in this paper. We formulate the kernel UDP theory and develop a two-stage method to extract kernel UDP features: namely weighted Kernel PCA plus UDP. The experimental results on the FERET and ORL databases show that the proposed kernel UDP is effective.

A framework for web-based intelligent decision support enterprise

This paper presents the agent-enabled communication mechanisms in designing decentralized distributed decision support enterprise where each decision-support service center is self-contained and can function independently, at the same time are all connected in an Internet-like enterprise-wide network environment for cooperative problem solving. Comparing to the traditional client/server paradigm, this approach provides a more efficient and secure solution since using intelligent agents leads to reduction of network traffic while protecting private information from being exposed In addition, with the collaboration of the emerging mobile agents concept and the distributed object-oriented computing mechanisms, these physically distributed decision centers can be logically unified to provide a better mode of communication with increased efficiency and flexibility, as well as improved re-usability of distributed components. A prototype system is developed using IBMs Aglets and Java/spl trade/ RMI.