Treating the information retrieval (IR) task as one of classification has been shown to be the most effective way to achieve high performance. In real-world Systems, a human is the ultimate determinant of relevance and must be integrated symbiotically into the control structures. We report on a hybrid, Human-Assisted Computer Classification system that opportunistically pairs processes of Active Learning and User Modeling to produce a high-Q computational engine. Top-down human goals are impedance-matched with bottom-up corpus analysis utilizing critical control loops. The System contributions of humans and machines as 'Proxy,' 'Assessor,' and 'Classifier' elements are blended through inter-related 'Model,' 'Match,' and 'Measure' processes (M3) to achieve consistently high precision IR with high recall. We report results for over a dozen topics, with confirmation of internal measures from topic 103 of the 2008 TREC legal track's interactive task.

The shared use of artifacts is, we argue, supported by latent border resources, which lie beyond what is usually recognized as the canonical artifact. These unnoticed resources are developed over time as artifacts are integrated into ongoing practice and stable conventions or genres grow up around them. For a couple of reasons, these resources may now deserve increased attention. First, because they lie outside conventional frames of reference, many new designs and design strategies inadvertently threaten to remove resources on which users rely. Second, because of the increasingly rapid proliferation of new technologies, users have less time to develop new border resources. Consequently, we suggest, designers now need to understand more fully the role border resources play and to work more directly to help users develop them. Meeting these goals will require more than an intensification of user-centered design. It will require a fundamental redirection of the way many designers look at both artifacts and users.

(1994). Patrolling the Border: A Reply. Human–Computer Interaction: Vol. 9, Context in Design, pp. 137-149.

No abstract available.

The purpose of this symposium is three-fold:

Much of the work in the field of computer human interaction consists of finding out what is wrong with existing interfaces or which of several existing alternatives is better. Over the next few decades, the possibilities for computer human interaction will explode. This will be due to: 1) continued decrease in the costs of processing and memory, 2) new technologies being invented and existing technologies (e.g., handwriting recognition, speech synthesis) being extended, 3) new applications and 4) new ideas about how people can interact with computers.

Although most technological efforts to enhance creativity have focused on constructing empowering tools for the individual, such efforts fail to do justice to the fact that much of our creativity is a social phenomenon involving the rubbing of prearticulate ideas against each other in a supportive manner. We thus need tools that help pull our thoughts from chaos to order and that enable us, collaborating with ourselves or others, to suspend judgment and let each idea be a trigger and a backdrop for the next idea.

Research on human factors in computer systems has emphasized supporting individuals. This panel will discuss new issues that emerge when computer systems support groups of people and whole organizations. Malone (see following paper) will suggest a broadening of the definition of user interfaces to include “organizational interfaces” and will indicate how a theoretical base for such an endeavor might be developed. Then Cashman will describe a “coordinator tool” in use at DEC for tracking the assignment of tasks to people in activities such as software maintenance. Finally, Brown will suggest how computer systems can be designed to radically increase the bandwidth of cooperation in groups by, for example, exploiting linguistic notions of context.

Intelligibility is one of the key factors affecting the acceptance and effective use of information systems. In this article, we discuss the ways in which recognition of this factor challenges current system design strategies, as well as current theoretical perspectives and research methodologies. In particular, we claim that in order to understand the problem of system intelligibility, we must focus not only on the cognitive, but also on the social aspects of system use. After considering some of the sources of users' difficulty in understanding information systems, we propose a new global philosophy for interface design: design for the management of trouble. We discuss the design implications of four mechanisms for improving system intelligibility: (1) useful mental models of the system and its associated subsystems, (2) communicative repair in user-system interaction, (3) new training strategies, and (4) use of the larger social environment as an aid to understanding information systems. We then discuss the possibility of developing intelligent systems capable of providing assistance and feedback related specifically to users' actions. We claim that development of such systems requires understanding the mechanisms for achieving mutual intelligibility in interaction and propose new research approaches for investigating these mechanisms. In the final section, we elaborate on the relationship between information systems and the larger social environment, suggesting that the functionality and design of information systems can deeply influence the surrounding culture. We propose adopting a goal of socially proactive design and discuss the possibilities for embedding new paradigms for communication and problem solving in specialized information systems.

This paper describes a pragmatic question-answering system which was developed for experimental use in a computer-assisted instruction environment. The two primary design goals of this system are 1) the representation of dynamic processes along with their causal interrelationships, and 2) the generation of responses in units of discourse greater than simple sentences. A method of modeling processes with augmented finite-state automata is discussed and is shown to permit efficient inferencing as well as provide a satisfactory deep structure for paragraph generation. Coupled with the dynamic process model is a semantic/conceptual network which contains static or factual information. Examples of the inferencing techniques using both the automaton model and the semantic network are given.