Michelle X. Zhou

Visual task characterization for automated visual discourse synthesis

To develop a comprehensive and systematic approach to the automated design of visual discourse, we introduce a visual task taxonomy that interfaces high-level presentation intents with low-level visual techniques. In our approach, visual tasks describe presentation intents through their visual accomplishments, and suggest desired visual techniques through their visual implications. Therefore, we can characterize visual tasks by their visual accomplishments and implications. Through this characterization, visual tasks can guide the visual discourse synthesis process by specifying what presentation intents can be achieved and how to achieve them.

Negotiation for automated generation of temporal multimedia presentations

Creating high-quality multimedia presentations requires much skill, time, and effort. This is particularly true when temporal media, such as speech and animation, are involved. We describe the design and implementation of a knowledge-based system that generates customized temporal multimedia presentations. We provide an overview of the system’s architecture, and explain how speech, written text, and graphics are generated and coordinated. Our emphasis is on how temporal media are coordinated by the system through a multi-stage negotiation process. In negotiation, media-specific generation components interact with a novel coordination component that solves temporal constraints provided by the generators. We illustrate our work with a set of examples generated by the system in a testbed application intended to update hospital caregivers on the status of patients who have undergone a cardiac bypass operation.

Data characterization for automatically visualizing heterogeneous information

Automated graphical generation systems should be able to design effective presentations for heterogeneous (quantitative and qualitative) information in static or interactive environments. When building such a system, it is important to thoroughly understand the presentation-related characteristics of domain-specific information. We define a data-analysis taxonomy that can be used to characterize heterogeneous information. In addition to capturing the presentation-related properties of data, our characterization takes into account the users information-seeking goals and visual-interpretation preferences. We use automatically-generated examples from two different application domains to demonstrate the coverage of the proposed taxonomy and its utility for selecting effective graphical techniques.

Top-down hierarchical planning of coherent visual discourse

A visual discourse is a series of connected visual displays. A coherent visual discourse requires smooth transitions between displays, consistent design within and across displays, and successful integration of new information into existing displays. We present an approach for automatically designing a coherent visual discourse. A top-down, hierarchical-decomposition partial-order planner is used to efficiently plan the visual discourse. Visual representations are modelled as visual objects, graphical techniques are employed as planning operators, and design policies are encoded as constraints. This approach not only improves the computational efficiency compared to search-based approaches, but also facilitates knowledge encoding, and ensures global coherency.

Managing networks through a virtual world

As network services become more sophisticated to handle multimedia exchange, demands for real-time network control are increasing. With 2D workstations, managers can observe network activities only at a distance. Direct observation requires a 3D interface combined with a structured control architecture. >

Management of broadband networks using 3D virtual world

Just as broadband networks will enable user-to-user communications to extend from textural services to those employing multimedia, they will also enable a management environment that can take advantage of increased bandwidth and multimedia technology. The fundamental advances incorporated in such an environment can provide efficient solutions to the problem of information management. To establish this environment, the authors tackle the fundamental problems of observability and controllability of broadband networks. A virtual world provides a next-generation network management interface through which a user can observe and interact with the network directly in real time. The system that the authors are developing uses a 3D virtual world as the user interface for managing a large gigabit ATM network. It provides the capability for experimentation in all aspects of network transport, control and management.<<ETX>>

Automated Visual Presentation: From Heterogeneous Information to Coherent Visual Discourse

Automated visual presentation systems should be able to design effective presentations for heterogeneous (quantitative and qualitative) information. They should also be able to work in static or interactive environments and capable of employing a wide range of visual media and visual techniques. In this paper, we focus on three tasks in building visual production systems: establishing a thorough understanding of the presentation-related characteristics of domain-specific information; classifying several types of visual information and capturing their distinct syntactic, semantic, and pragmatic features; and formulating a set of design principles.We define a data-analysis taxonomy to characterize heterogeneous information. In addition, we have modeled presentation context information such as audience identity to produce user-centered visual design. To utilize and manipulate visual information, we have classified it into visual objects and visual tools based on its role in the visual production process. To guide the visual design process, we have formulated a set of design principles that ensure the expressiveness and effectiveness of a design. To test and evaluate our work, we have developed a prototype system called IMPROVISE based on the research results. We use examples generated by IMPROVISE to illustrate how it constructs visual presentations.

A virtual world for network management

Existing network management systems typically use a combination of textual displays and 2-D directed graph representations of network topology. A network management system is being designed that uses a virtual world presented through a 3-D stereo display and manipulated with a 3-D mouse. The goal is to allow the user to better understand and control the structure and behavior of a large, complex network. In the current prototype, the user interacts with a 3-D representation of a network whose topology and behavior are specified by a separate network emulator. The user can choose from among a set of different views of the network. For example, one view shows a selected virtual path as a series of logical links contained within a physical path. The system will serve as a testbed for the knowledge-based design of network visualizations.<<ETX>>

Efficiently planning coherent visual discourse

A visual discourse is a series of connected visual displays. A coherent visual discourse is characterized by smooth transitions between displays, consistent design within and across displays, and successful integration of new information into existing displays. We use a topdown, hierarchical-decomposition, partial order planner to efficiently construct a visual discourse from scratch, taking advantage of parametrized primitive visual objects that serve as building blocks in the design process. Visual representations are modeled as visual objects, graphical techniques are employed as planning operators, and design policies are encoded as constraints. This approach not only improves computational efficiency compared to search-based approaches, but also facilitates knowledge encoding, and ensures global coherency.

Automated visual discourse synthesis: coherence, versatility, and interactivity

In this proposal, we present comprehensive and systematic approaches of building systems that can automatically generate coherent visual discourse for interactive envirornments. A visual discourse refers to a series of connected visual displays. A coherent visual discourse requires smooth transitions between displays, consistent designs within and among displays, and effective integration of various components. Our research focuses in part on establishing a general framework by abstracting various generation systems and providing a reference model in which a specific system is considered an instantiation of the framework. In other words, any automated graphics generation system must contain a knowledge base, an inference engine, a visual realizer and an interaction handler. As a consequence, not only can a general framework serve as a template from which a specific generation system can be instantiated, but the framework also can be used as a base for comparing or evaluating different systems. We concentrate on the basic issues involved in establishing these four core components. In particular, we identify various knowledge sources and determine effective knowledge representation paradigms in constructing the knowledge base. We emphasize the efficiency, usability, and flexibility issues in modeling the inference engine. We are concerned with portability and parallelization issues in building the visual realizer, and we also take into account interaction capabilities for interactive environments. To demonstrate the generality and comprehensiveness of the framework, we address its application to the design of coherent visual discourse for heterogeneous information in interactive environments. Within such discussions, heterogeneous information refers to both quantitative and qualitative, or static and dynamic information. In addition, we assume that the system aims to support a wide variety of visual techniques, ranging from individual 2D displays to interactive 3D animation sequences. We describe a system called IMPROVISE (Illustrative Metaphor Production in Reactive Object-oriented VISual Environments) that serves as a proof-of-concept prototype. IMPROVISE is built based on our framework, aiming to automatically generate coherent visual discourse for various application domains in interactive environments. IMPROVISE has been used in two testbed application domains to demonstrate its generality and flexibility. Examples from both domains will be given to illustrate IMPROVISE’s generation process and to identify the future research areas.