As Internet access via cell phone browsers becomes more common, new website interfaces are emerging. Hierarchical menus are effective for such interfaces due to their compact presentation. For menus with many levels and possible destinations, personalized menu organizations can potentially expedite navigation. This thesis contains an empirical and analytical evaluation of two menu personalization techniques based on user preferences, Expanded menus and Expanded/Reordered menus. A 30 participant experiment revealed that both types of personalized menus resulted in faster performance and higher evaluations from the users, with the Expanded/Reordered menu ranking the highest in both respects. One interesting finding was that individual users' evaluation was not consistent with their performance. Data collected from the performance study was used to evaluate an existing GOMS model of cell phone menu traversal, which demonstrated significant limitations in the generality of the model. However, the results also suggest that individual differences in user performance may account for much of the error in model predictions.

SILO is a new internetworking architecture that represents a significant departure from current philosophy and practice. The architecture consists of building blocks of fine-grain functionality, explicit support for combining elemental blocks to accomplish highly configurable complex communication tasks, and control elements to facilitate (what is currently referred to as) cross-layer interactions. It takes a holistic view of network design, allowing applications to work synergistically with the network architecture and physical layers so as to meet the application's needs within resource availability constraints. The SILO research advocates a non-layered architecture based on silos of services assembled on demand and specific to an application and network environment. With the goal to facilitate what in today's layered architecture is referred to as ``cross-laye' interactions, in a manner that meets the exact user requirements and optimizes performance, the main focus of this research work is on developing mechanisms to optimize the construction of SILOs (stack of services) in a manner that takes into account service specific constraints, current network conditions and user policies.

Wavelength division multiplexing (WDM) has emerged as a viable solution to the increasing bandwidth demands of current backbone networks. Traffic grooming allows the resources available on a single wavelength to be utilized by multiple traffic streams. These networks will support large amounts of traffic from various sources. In view of this, our research focuses on the analysis of such types of networks, and in particular on a traffic-groomed tandem optical network that supports multiple sources each with unique resource requirements. We developed a decomposition algorithm for the analysis of this network using a modified version of Courtois? method. This work was extended to analyze tandem networks with a large number of wavelengths and resources per wavelength using a modified version of the recursion developed by Nilsson et. al. Finally, we extended our work to a traffic-groomed tandem optical network employing alternate routing and supporting traffic from multiple sources. For each method, numerical results show that, overall, our approximation provides good accuracy.

Online auctions provide a very competitive market for a variety of goods and services and have gained a lot of popularity over the last few years. The increasing participation of buyers and sellers in these auctions has triggered the development of software bidding agents. These agents are automated agents that are designed to place bids on behalf of the user. In recent years, simulation of online auction markets has become a major research area in the field of electronic commerce. This is because testing different bidding strategies in a real auction environment involves risks and can result in heavy losses. Moreover, testing in a real life auction involves substantial amount of time and participation, in order to derive a definite conclusion about the bidding agent?s performance. We propose to solve this problem by designing and implementing a tool for simulating online auction markets. Our simulation tool called <i>fauxBay</i> can be configured to develop test scenarios for testing bidding agent strategies. Currently, <i>fauxBay</i> supports the reserve price auction mechanism found on eBay. However, support for other auction mechanisms can be easily integrated with <i>fauxBay</i> in the future. We have designed <i>fauxBay</i> to offer two modes of simulating auction environments for testing bidding strategies: event-based and clock-based. Event-based simulation is used to obtain a quicker analysis of the bidding agent?s performance, by moving forward to each event rather than moving linearly by the clock. In contrast, the clock-based simulation offers to test and gauge a bidding agent?s performance in an auction environment where time moves linearly by the clock similar to real life online auctions. A key characteristic of <i>fauxBay</i> is that it tests bidding agent?s strategies against real eBay data on completed auctions. Therefore, the results obtained by testing bidding strategy in <i>fauxBay</i> give a better insight into its anticipated performance in a real life online auction. Overall, <i>fauxBay</i> is designed to be extensible, configurable and give close to realistic results on the performance of bidding agents.

This work describes an experiment comparing knowledge sharing in two distinct team structures. Given the increasing importance of knowledge in the workplace, especially software engineering, we are interested in paradigms that can assist in knowledge management. To this end, we conducted an experiment to determine how the paired or solo programming model affects knowledge sharing during a software project. We show a pattern of results that suggest pair programming has a positive effect on knowledge sharing. We also find development time is somewhat higher when using paired programming, but product quality is unaffected by programming method though these results are not statistically significant.

Interactive narrative environments have been the focus of increasing attention in recent years. A key challenge posed by these environments is narrative planning, in which a director agent orchestrates all of the events in an interactive virtual world. To create effective interactions, the director agent must cope with the task's inherent uncertainty, including uncertainty about the user's intentions. Moreover, director agents must be efficient so they can operate in real time. To address these issues, we present U-DIRECTOR, a decision-theoretic narrative planning architecture that dynamically models narrative objectives (e.g., plot progress, narrative flow), storyworld state (e.g., physical state, plot focus), and user state (e.g., goals, beliefs) with a dynamic decision network (DDN) that continually selects storyworld actions to maximize narrative utility on an ongoing basis. DDNs extend decision networks by introducing the ability to model attributes whose values change over time; decision networks extend Bayesian networks by supporting utility-based rational decision making. The U-DIRECTOR architecture also employs an n-gram goal recognition model that exploits knowledge of narrative structure to recognize users' goals and an HTN planner that operates in two coordinated planning spaces to integrate narrative and tutorial planning. U-DIRECTOR has been implemented in a narrative planner for an interactive narrative learning environment in the domain of microbiology in which a user plays the role of a medical detective solving a science mystery. Formal evaluations suggest that the U-DIRECTOR architecture satisfies the real-time constraints of interactive narrative environments and creates engaging experiences.

Recent advances in computer graphics and multimedia technologies have contributed to the development of interactive media-rich systems that generate media elements dynamically in multiple modalities to present information in an effective and appealing manner to users. Utilizing media elements in an application that requires them to be temporally coordinated in real-time presents synchronization issues because the wrong timing or duration of media elements will be easily detected by the user. Most media-rich systems have solved various aspects of the media element synchronization problem. However, they have done so in an ad hoc manner without providing a generic reusable framework. This thesis proposes a framework for real-time synchronization in intelligent media generators. Our research addresses six primary issues that are essential to creating media-rich systems that do not utilize an ad hoc solution to their media element synchronization problems: synchronization of media elements, decoupling high-level and low-level processing, extensibility, media element sequencing, real-time performance, and domain-independence. The framework has been implemented in an agent-based multimedia generator for an intelligent tutoring system to demonstrate its feasibility.

Advances in microprocessors, memory, and radio technology have enabled the emergence of embedded systems that rely on communication systems to exchange information and coordinate their activity in spatially distributed applications. Developing embedded communication systems that are efficient and reliable, is a challenge due to the trade-offs imposed by the conflicts between application requirements and hardware constraints. In this thesis, we present RaPTEX, an integrated development environment (IDE) for embedded communication systems. RaPTEX consists of three major subsystems: a graphical module to facilitate component composition, code generation with access to component-level parameters, and a performance evaluation framework for allowing system designers to explore what-if scenarios and clearly expose the trade-offs of their choices. We also present two case studies of developing wireless sensor network applications using RaPTEX.

Tool use is a hallmark of human intelligence, which has not fully been explored in the artificial intelligence research community. Research in cognitive neuroscience on primates suggests that not only do we maintain a mental representation of our body, but the body schema is modified to include a tool during intentional tool use (Iriki et al., 1996). We have developed a habile (tool-using) agent, based on the Sony Aibo platform, that can pick up a stick and use it as a tool to reach objects previously out of its range. The agent uses a recurrent neural network developed by Steinkühler and Cruse (1998) for maintaining an internal body schema used to find appropriate postures for reaching and grasping tools. We argue that analysis of activities of such tool using agents offers an informative way to evaluate intelligence.

In artificial intelligence (AI) research, embodied systems have received increasing attention since the 1990s. How to bridge the gap between raw sensorimotor data and symbolic representation in a robotic agent is still an open question. The research described in this document is inspired by theories in cognitive science, such as concept theory and embodied realism, as well as work in robotics and AI. The general goal of this research is to build a system capable of acquiring and maintaining semantic knowledge for higher-level reasoning, in particular reasoning about the use of tools, from the embodied experience of a cognitive agent in a simulated environment or in the real world. This research addresses cognitive theories of embodiment, the design of a general computational architecture, and the design and implementation of AI techniques for solving tool-using problems. One of the major contributions of this research is to provide a computational architecture for an embodied agent that can capture semantic relations from its interactions with the world, sufficient to support effective tool use both in short-term predictions and plan generation. As a result, we have implemented an example of this architecture in an Action Schema Generator, or ASG, which can automatically generate production rules and symbolic representations from a simulated agentâs embodied experience without losing the capability of transferring the knowledge backwards to its original numerical sensorimotor format. We have developed pragmatic methods to evaluate the performance of ASG, at the component level and the system level, in simulated and real scenarios, for tasks with and without tools. We also have compared our design with other robotics and cognitive architectures, including behavior-based robotics, Neuroevolution, and psychologically inspired architectures. We believe that our work can provide a general foundation for embodied agents, and should be useful in future research.