Vol.7 No.2 June 15, 2008       

Research Articles:
Reactivity-based Approaches To Improve Web System's QoS (pp093-132)
        A.C.M. Pereira, L.A. Silva, W. Meira Jr., and W.D.S. Filho
Understanding the characteristics of Internet services workloads is a crucial step to improve the Quality of Service (QoS) offered to Web users. Moreover, studying and modeling the user behavior is important to analyze the performance and the scalability of web servers. This knowledge may be used, for instance, to build workload generators that help evaluating the performance of those servers. Current workload generators are typically memory-less, being unable to mimic actual user interaction with the system. As the basis of this work, we propose a hierarchical characterization and simulation model focused on the user behavior, named USAR.   In fact, there is strong evidence that a significant part of the user behavior depends on its satisfaction. Users reactions may affect the load of a server, establishing successive interactions where the user behavior affects the system behavior and vice-versa. It is important to understand this interactive process to design systems more suited to user requirements. In fact, the user reactivity, that is, how the users react to variable server response time, is usually neglected during performance evaluation. In this work we study and explain how this reactive interaction is performed by users and how it affects the system's performance.   Web applications demand requirements, such as performance and scalability, in order to guarantee QoS to users. Due to these requirements, QoS has become a special topic of interest and many mechanisms to provide it have been proposed. In this work, we address the use of reactivity to improve Internet services. We propose and evaluate new admission control and scheduling mechanisms. We designed and implemented the USAR-QoS simulator that allows the evaluation of the new strategies considering the dynamic interaction between client and server sides in Internet services. We simulate the new strategies using a TPC-W-based workload. The experiments show the benefits of the reactive policies which can result in better QoS for Internet Services, improving the user satisfaction. We also propose a hybrid admission control and scheduling mechanism that combines both reactive approaches. The results show benefits in terms of response time and user satisfaction.

Obstacles Reveal the Needs of Mobile Internet Services (pp133-157)
        M. Sasajima, Y. Kitamura, T. Naganuma, K. Fujii, S. Kurakake, and R. Mizoguchi
Growth in the mobile internet services industry has seen a marked increase in the number of mobile internet services provided, and this has made proper structuring and organization of the services difficult. Present methods of service provision have proven insufficient to guide users efficiently to the services they need. To solve this problem, a task-oriented menu, which enables users to search for services by what they want to do and not by category, has been proposed. Construction of the task-oriented menu is based on a task ontology modeling method which supports descriptions of user activities, such as task execution and defeating obstacles encountered during the task. This paper discusses a task ontology-based modeling method which supports descriptions of users’ activities and related knowledge, such as how to solve problems that the users encounter and how to prevent or solve them on the spot. Models described by our method contribute to designing, testing, and improving mobile internet services.

Semantic spam filtering from personalized ontologies (pp158-176)
        V. Eyharabide and A. Amandi
One of the biggest problems that Internet faces is the increase of email spam. The main drawback with previous anti-spam filters is that they are based only on 1) the syntactical features of words lacking semantic analysis, or 2) on what the majority of users regard as spam without considering the individual preferences of a particular user. In this paper we present a spam email filter that personalizes its filtering process using an email user profile that contains the user’s preferences regarding emails. Our innovative email user profile is based not only on some common user profiling techniques but also on the knowledge contained in a domain ontology. The user profile is used to learn which spam emails (although unsolicited and large-scale sent) are interesting for the user, despite they are spam. The encouraging experimental results provide empirical evidence of the effectiveness of using an ontological approach to user profiling in an email spam filter.

Back to JWE Online Front Page