Selecting transparently a proper network connection for voice communication will be a fundamental requirement in future multi-mode heterogeneous wireless network. This paper extends always best connected (ABC) to a fine-grain paradigm called always best network connection (ABNC) to address this issue. Instead of selecting a best access network as in conventional ABC, ABNC enable users to select a best network connection, which consists of source and destination access network pair, to satisfy quality constraint and users' preference. To support ABNC, we develop a user profile to specify network connection priority. Meanwhile we extend SIP and propose a network selection information service (NSIS) based on MIH (media independent handover) to collect information of both source and destination access networks for decision making. Finally, analytic hierarchy process (AHP) is used to recommend a network connection with assistance of user profile and NSIS. An example is illustrated to show that AHP can successfully select a good network connection that fulfill the requirement of users.

Intention is an important factor to drive human behavior. With a certain intention, an action becomes more natural and reasonable. In this paper, an agent with intention, which not only possesses facial expressions, hand gestures, body motions, but also holds memory and personality, is presented. The agent exploits the social dialogue to show intention and its care for students in a learning system. The agent will purposefully guide students in the learning process, care about their needs during learning, provide empathy feedback in order to lessen their negative emotions, and offer appropriate reproach or encouragement to help them overcome the obstacles to English learning. A preliminary experiment will be done to show that this agent can interact with students more naturally and meanwhile strengthen their will to learn as well as make them more willing to accept the opinions from the agent.

This study proposes the application of the MCSCL groupware for the hearing-impaired students to engage in learning the mathematics word problems. This groupware, which integrates the collaborative argumentation theory, the scaffolding tools, and the comic, is to be applied in the activities of the collaborative learning. We present the system architecture which describes the implementation based on integration of the PDA, the network technology, and the learning strategy. Furthermore, this study employed the groupware as a channel for the hearing-impaired students to improve the experience of learning the mathematics word problems.

The Universal Mobile Telecommunication System (UMTS) adopts WCDMA as the air interface that provides variable transmission rate services. Four traffic classes are identified in UMTS, given as controversial, streaming, interactive, and background classes. To efficiently utilize the radio bandwidth, the shared channel technology is proposed to deliver interactive and background traffics. This paper proposes a "shared-channel assignment and scheduling" (SCAS) algorithm to allocate shared channels to interactive and background connections. We formally prove that the SCAS algorithm can satisfy the transmission rate requirement of each interactive connection and can efficiently utilize the radio bandwidth.

We utilize Web and database technology to construct a journal based portfolios assessment system to realize a legitimate peripheral participation (LPP) (J. Lave et al., 1990) learning community. In the system, we divide system into three layers, the view layer presents data by the Oracle Portal Server, the data layer stores data by the Oracle Database Server and the operating control layer operates workflow by the Oracle Database trigger. Job data analysis and a questionnaire indicates that the Web Journal is effective to promote knowledge exchange. Students can improve learning performance and it can form a LPP structure. The purpose of using the journal approach [Roth, W.M. (1997)] is to emulate the forming of academic societies formed by journals. At the same time, the achievement of students in a course is based on their publications in journals and the positions they played in the journals. First, students submit their work such as homework, learning portfolios, reports of organizing contents in the discussion, learning problems, related Web sites. Each submitted article would be assessed by three peers. Then the reviewers will review the submitted works. The reviewers will give comments to students. The journal system will publish good articles. Articles published include submitted works and reviewer's comments hoping that students will contribute and exchange their knowledge.

The mesh-connected multiprocessor has become popular because of its simple and regular structure. A new data structure, the R-array, is proposed to represent the mesh at first. The element in the R-array stores statistical information about occupied conditions of the mesh. Statistical information of the R-array can direct the allocation process to jump to the processes that can serve as a base of a free submesh. Based on a simple and reasonable assumption, we develop a stochastic process to analyze behaviors of the proposed scheme. The proposed scheme is the first whose probabilities of locating free submeshes under different workloads are precisely computed. These results can be applied to each full-recognition scheme. In addition, the execution costs of the proposed scheme can also be accurately calculated. Finally, simulations are performed which show that the proposed schemes are faster than most.

The mining of a complete set of frequent itemsets will lead to a huge number of itemsets. Fortunately, this problem can be reduced to the mining of frequent closed itemsets (FCIs), which results in a much smaller number of itemsets. The approaches to mining frequent closed itemsets can be categorized into two groups: those with candidate generation and those without. In this paper, we propose an approach to mining frequent closed itemsets without candidate generation with a data structure called the frequent pattern list (FPL). We designed the algorithm FPLCI-mining to mine the FCIs. Experimental results show that our method is faster than previous ones.

An ad hoc network is formed by a group of mobile hosts communicating over wireless channels. Such a network operates without any fixed network interaction and centralized administration. For the network environment is highly dynamic, the established routing paths are so fragile to maintain stable communication, especially the connection-oriented communications are quite difficult to be maintained. In our approach, the network infrastructure is initially constructed by several communication groups, which are called the dynamic group infrastructure. This paper proposes the group-level connection maintenance based on the dynamic group infrastructure to implement connection-oriented routing. Our approach enhances the routing robustness of connection-oriented communications. The simulation results show that this routing approach under such a construction outperforms the previous works in message complexity for connection-oriented communications.

The World Wide Web is a world of great richness, but finding information on the Web is also a great challenge. Keyword-based querying has been an immediate and efficient way to specify and retrieve related information that the user inquires. However, conventional document ranking based on an automatic assessment of document relevance to the query may not be the best approach when little information is given, as in most cases. In order to clarify the ambiguity of the short queries given by users, we propose the idea of concept-based relevance feedback for Web information retrieval. The idea is to have users give two to three times more feedback in the same amount of time that would be required to give feedback for conventional feedback mechanisms. Under this design principle, we apply clustering techniques to the initial search results to provide concept-based browsing. We show the performance of various feedback interface designs and compare their pros and cons. We measure precision and relative recall to show how clustering improves performance over conventional similarity ranking and, most importantly, we show how the assistance of concept-based presentation reduces browsing labor.

The explosion of the World Wide Web as a global information network brings with it a number of related challenges for information retrieval and automation. The link structure, which is the main feature of the hypermedia environment, can be a rich source of information for exploration. This paper is centered around the exploitation of hyperlinks in the subject of automatic discovery. The paper discusses the design of site traversal algorithms and identifies some important parameters for hyperlink-based information discovery on the Web. Meanwhile, it proposes a new evaluation method, the search depth, for comparing various hyperlink traversal algorithms.

The current search tools for retrieving information on the WWW are not suitable for building customized information repositories because these search tools are designed for general users with the result of only an unstructured collection of documents. We present a personal digital library capable of efficiently retrieving information on the World Wide Web. We adopt several new strategies to overcome the shortcomings of current tools. The first strategy, classification, merges and organizes the retrieved documents to put them in a structural hierarchical frame. The second strategy, user profile, saves time and bandwidth for the access of the documents and permits the users to build their own customized category structure. The third strategy, multi-search, capitalizes on the power of multiple search engines to broaden the domains of information sources and alleviate the overloading of a single search engine. Furthermore we derive in detail the techniques for speeding up the iterative process of clustering.

Simulated annealing can be viewed as a process that generates a sequence of Markov chains, i.e., it keeps no memory about the states visited in the past of the process. This property makes simulated annealing time-consuming in exploring needless states and difficult in controlling the temperature and transition number. In this paper, we propose a new annealing model with memory that records important information about the states visited in the past. After mapping applications onto a physical system containing particles with discrete states, the new annealing method systematically explores the configuration space, learns the energy information of it, and converges to a well-optimized state. Such energy information is encoded in a learning scheme. The scheme generates states distributed in Boltzmann-style probability according to the energy information recorded in it. Moreover, with the assistance of the learning scheme, controlling over the annealing process become simple and deterministic. From qualitative and quantitative analyses in this paper, we can see that this convenient framework provides an efficient technique for combinatorial optimization problems and good confidence in the solution quality.

We describe a practical train scheduling system (called TSS) designed for the Taiwan Railway Bureau (TRB). TSS has two major components. The first is Auto Scheduler which includes an initial scheduler, a conflict finder, and a conflict resolver. Auto Scheduler uses the concept of bugging as problem solving tactics. The second is Manual Scheduler which provides some effective editing functions for users to tune schedules generated by Auto Scheduler. Through an easy-to-use user interface, these components can be used to solve large-scale complex train scheduling problems.>

A stochastic approach called the annealing-genetic algorithm is presented for solving some well-known combinatorial optimization problems. This approach incorporates genetic algorithms into simulated annealing to improve the performance of simulated annealing. The authors' approach can be viewed as a simulated annealing algorithm with population-based state transition and with genetic-operator-based quasi-equilibrium control and as a genetic algorithm with the Boltzmann-type selection operator. The goals of efficiency in the algorithm are (1) the gap between final solution and the optimal solution should be around 3% or less, and (2) the computation time should be bounded by a polynomial function of the problem size. Empirically, the error rate of the proposed annealing-genetic algorithm for solving the multiconstraint zero-one knapsack problem is less than 1%, for solving the set partitioning problem is less than 0.1%, and for solving the traveling salesman problem is around 3%.>