Elise H. Turner

Organization and reorganization of autonomous oceanographic sampling networks

Systems such as autonomous oceanographic sampling networks (AOSNs) that have multiple autonomous or semi-autonomous components must have an organization which specifies the interactions between the components to allow them to distribute and accomplish the systems tasks. AOSNs present a special challenge. They will be deployed for long periods of time, and they are open systems whose composition will change over time. Such systems require the ability to autonomously organize and reorganize in response to changes in its composition, the environment, or the mission. In this paper, we present preliminary results from a project whose goal is to develop mechanisms to allow AOSNs to self-organize and reorganize. We discuss characteristics of AOSNs which impact their organization and give an overview of an approach which addresses their special requirements. We discuss a simulation methodology designed to simulate the aggregate properties of the protocols developed, and we describe preliminary results obtained using that simulator.

Retaining majors through the introductory sequence

Retention is an important issue for Computer Science Departments. In many cases students leave the major due to frustrations with programming in the complex languages often used in CS1 and CS2 or because they do not understand that computer science is much more than programming. We have redesigned our introductory sequence to include a rigorous, non-programming introduction to the field and a CS1 course which uses Scheme so that students can focus on the principles of programming instead of the complexities of a particular language. In the first year that we have required these courses in our major, we have had positive results. In this paper, we describe what we have learned through discussions with students and student surveys.

Teaching entering students to think like computer scientists

This paper describes a new course developed at University of Maine to help students better understand the discipline of computer science and to aid us in recruiting and retaining majors. The course verview of computer science, but also, through focusing on particular topics at an advanced level, begins to teach students how computer scientists think about problems. The course has been taught in Fall 2002, 2003 and 2004. This paper describes the course and discusses our results from the first two years.

Using Explicit, A Priori Contextual Knowledge in an Intelligent Web Search Agent

The development of intelligent Web search agents will become increasingly important as the amount of information on the Web continues to grow. Intelligently searching the Web depends on the searcher understanding not only the context of the query, including the person for whom the search is being done, but also the context of the results, including the information sources and the retrieved information itself. Consequently, intelligent Web search agents will need to have mechanisms for representing and using contextual knowledge. In this paper, we discuss the kinds of contexts and contextual knowledge such an agent will encounter. We use as an example a Web search agent we are beginning to develop, ferret, that will search for scholarly information about music. We then propose some ways in which explicitly represented, a priori contextual knowledge can be used by the search agent, and we discuss directions for future research.

Determining the usefulness of information from its use during problem solving

Information is an impor tant resource tha t should be shared by collaborating agents. Yet, communication can be costly and agents often know much more than they can say. Agents, therefore, should choose information to communicate tha t will be the most beneficial to others. In this paper, we present a new approach to selecting information to communicate. Our approach relies on an agent s use of information in its own problem solving to suggest how useful the informat ion may be to others. Consequently, it requires no additional knowledge and little computat ional effort beyond tha t needed by the agent to reason about the problem solving task. Heuristics t ranslate problem solving experience into usefulness ratings for information. When its usefulness rating exceeds a threshold, the information is communicated.

A Unified Long-Term Memory System

Memory-based reasoning systems are a class of reasoners that derive solutions to new problems based on past experiences. Such reasoners use a long-term memory (LTM) to act as a knowledge base of these past experiences, which may be represented by such things as specific events (i.e. cases), plans, scripts, etc. This paper describes a Unified Long-Term Memory (ULTM) system, which is a dynamic, conceptual memory that was designed to be a general LTM capable of simultaneously supporting multiple intentional reasoning systems. Through a unique mixture of content-independent and domain-specific mechanisms, the ULTM is able to flexibly provide reasoners accurate and timely storage and recall of episodic memory structures. In addition, the ULTM provides support for recognizing opportunities to satisfy suspended goals, allowing reasoning systems to better cope with the unpredictability of dynamic real-world domains by helping them take advantage of unexpected events.

Aspects of Context for Understanding Multi-modal Communication

Context is important for AI applications that interact with users. This is true both for natural language interfaces as well as for multi-modal interfaces. In this paper, we consider aspects of context that are important in a multi-modal interface combining natural language and graphical input to describe locations. We have identified several aspects of contexts in our preliminary study. We describe them here and discuss plans for future work.

A Constraint-Based Approach to Assigning System Components to Tasks

In multi-component systems, individual components must be assigned to the tasks that they are to perform. In many applications, there are several possible task decompositions that could be used to achieve the task, and there are limited resources available throughout the system. We present a technique for making task assignments under these conditions. Constraint satisfaction is used to assign components to particular tasks. The task decomposition is selected using heuristics to suggest a decomposition for which an assignment can be found efficiently. We have applied our technique to the problem of task assignment in systems of underwater robots and instrument platforms working together to collect data in the ocean.

Understanding pronominal references to quantified expressions

Quantified expressions present an interesting case for understanding pronominal reference because the quantifier which appears in the expression affects how the pronoun will be understood. The quantifier determines which sets will be placed in focus and predicts the relation between the sentence containing the quantifier and the following sentence (Moxey and Sanford 1993a). This information is essential to understanding the pronominal reference. In this paper, we discuss how a natural language processing system can take advantage of this information to understand pronominal references to quantified expressions.

Representing the Graphics Context to Support Understanding Plural Anaphora in Multi-modal Interfaces

Previous communication provides important context for new communication in an interaction. In natural language interfaces, the discourse context represents and maintains information about what has been said before. When other modes of communication are also used, they must also contribute to the context. In this paper, we describe how information about the graphics can be represented and maintained in the graphics context. We are particularly interested in how the graphics context can be used to support finding referents for plural anaphora.