HCI Bibliography Home | HCI Conferences | HYPER Archive | Detailed Records | RefWorks | EndNote | Hide Abstracts
HYPER Tables of Contents: 87899191Z9393X93Y93Z969797X

Proceedings of ACM Hypertext'91 -- Posters 1991-12-15

Fullname:ACM Hypertext'91 Conference -- Posters
Editors:Gary Perlman
Location:San Antonio, Texas
Dates:1991-Dec-15 to 1991-Dec-18
Standard No:hcibib: HYPER91Z
IKON: Developing a Prototype Static Third-Order Hypermedia System BIBA 1
  Hans C. Arents; Walter F. L. Bogaerts
This contribution presents the first results of our efforts to design and develop a static knowledge-based hypermedia system, called IKON (Intelligent Knowledge Objects Navigator). The architecture of this prototype is based on the Model-Map-View-Praxis (MMVP) architecture we have proposed for what we have called third-order hypermedia systems, or hypermedia systems which not only represent their contents' semantics, but also actively manipulate it. The implementation of IKON is based on two key ideas: the use of semi-formal or semi-structured objects for representing the information contents, network nodes and links, and the use of a frame formalism for representing the semantics of the information, the node types and the link types. The presentation of the nodes and the navigation of the links is based on the link navigation through message passing mechanism introduced in the MMVP architecture. Although only part of IKON has been implemented, some of the fundamental concepts and ideas behind the MMVP.
RelType: Relaxed Typing for Intelligent Hypermedia Representations BIBA 2
  Dilip K. Barman
Hypertext is an ideal medium to flexibly capture information, but the very flexibility precludes meaningful semantic content. RelType is a relaxed typing scheme that models hypertext as an object-oriented knowledge representation medium, by allowing but not requiring link and node type specifications with inheritable behaviors.
   RelType allows tractable retrieval from flexibly and minimally structured information that normally would be expected only from more formal representations. Traditional link following, as well as intensional query and inference based navigation, are supported information traversal modes. Knowledge posited into a potentially large hypertext can thus be intelligently retrieved and interpreted. The approach represents a novel integration of hypertext with term classification KR to usably and tractably provide semantic inference at a user-controlled level of richness.
An Architecture for Wide Area Hypertext BIBAHTML 3
  Tim Berners-Lee
Information Retrieval meets HyperText in the WorldWideWeb (W3) architecture which provides wide area information access between heterogeneous platforms, using the two user metaphors of hypertext jump and index search.
   The architecture has allowed many existing hypertext systems and information bases to be incorporated as part of the web by gateway servers. These map a portion of the W3 address space onto the address space of the system. In this way for example, Thinking Machines' "WAIS" data, Digital's "VMS/HELP" data, the Technical University of Graz's "Hyper-G" hypertext system, Internet news, and CERN's "XFIND" documentation database and telephone directory may all be browsed in a continuum with new hypertext data. W3 unites these systems and achieves a practical information universe.
   The W3 client/server architecture relies on a formal notation for the name or address of a document. The address notation is open to accommodate new protocols and name spaces as they develop within the IR and networking communities.
   Browsers use FTP, and a new search and retrieve protocol allowing negotiation between client and server about the data formats which each supports. Each browser handles as a minimum plain text, and simple SGML hypertext. SGML is used for transferring hit lists resulting from index searches, as well as being an optional documentation format. Currently in use are a graphic hypertext editor (on a NeXT machine) and a line-mode browser.
Integrating Existing Documents with Hypertext in NASA's Shuttle Mission Control Center Environment BIBA 4
  Debra S. Bettis
NASA's Space Shuttle Program like many other technical programs of its magnitude is supported by a large volume of existing technical documents. These documents are not only diverse but also abundant. Management, maintenance, and retrieval of these documents is a challenging problem by itself but; relating and cross-referencing this wealth of information when it is all on a medium of paper is an even greater challenge. Designing software to manage this information without requiring that text be re-authored is the goal of our hypertext effort.
Focused and Relativistic Hypertext Usability Assessment: Finding the Sweet Spots in Hypertext Design Space BIBA 5
  Mark H. Chignell; J. Felix Valdez; Rhona Charron
The design space for hypertext is very large and it is difficult to assess the global usability of all the points in that space. Consequently, in this paper the problem of hypertext usability assessment is considered from the perspective of formative evaluation of specific hypertext components. After reviewing some of the existing literature on hypertext usability, we recommend a featural analysis of hypertext usability based on feature dependent assessment methods. We then describe specific methodologies that have been developed for testing the usability of different approaches to link and landmark selection.
   Rather than develop global measures of usability, we favor an analysis of different components of hypertext. Thus many different usability assessment methodologies may be developed for various components of hypertext design, including the issues of link creation and landmark selection considered in this paper. To illustrate this approach, we have described three such methods. The first method for assessing link quality is obvious, and can be applied without much difficulty to most hypertext systems. The other two methods provide an interesting contrast in approaching the problem of landmark evaluation. The PI approach is a well-defined laboratory task, but it is not integrated into the hypertext browsing task. In contrast, the LESS approach integrates landmark selection into browsing so that landmarks are selected in exactly the same way (except for using a different window) as regular links are traversed. We regard the three methods presented here as illustrative of the types of feature-based hypertext usability assessment tools that should be developed as we move towards more focused analysis of hypertext usability. It is argued that the methodology for hypertext usability assessment described in this paper can be extended to cover a range of hypertext features, and that focused usability assessment is critical for development of hypertext models and environments. It is anticipated that this approach will help to identify the sweet spots in hypertext design space.
Open Multimedia Telecommunication Needs Hypertext Techniques BIBA 6
  Ralf Cordes; Hauke Peyn; Thomas Toepperwien; Thomas Weidenfeller
First attempts in modelling and structuring multimedia telecommunication services not seem to be flexible as they are required as distributed hypermedia platforms, offering personalized links, different access pathes and retrieval techniques or individualized views. In our approach (partly supported by the European Commission within the Project RACE R1038 Multimedia Communication Processing and Representation) we have combined an object oriented structuring of multimedia information according to the MHEG proposal with flexibility of the DEXTER model serving as framework for an open link server. We distinguish between object classes for particles (information nodes) like audio, video, text, pages (composite information units), links etc. The implementation has been made on SUN4 machines using SunOS4.1, Open Windows 2.0, and C++. A distributed demonstrator will be presented on TELECOM91 in Geneva in October. This prototype connects a multimedia terminal (SUN4 based), a videophone terminal and a videoserver (videodisk player) via a private broadband switch offering an extended ISDN system with a 140Mbit/s channel for highspeed data transfer in STM technique. The concept of hypertext offers the suitable features which have to be combined with the structuring of information units and service components.
Content Map Design and Knowledge Structures with Hypertext and Traditional Text BIBA 7
  Diana Dee-Lucas; Jill H. Larkin
Two experiments compared learning from three texts -- a "structured" hypertext with a content map organizing the text units, an "unstructured" hypertext with a menu-like content map, and a traditional text on a computer. Hypertext facilitated recall compared to traditional text, but the unstructured hypertext produced a more fragmented knowledge structure when readers lacked specific study goals. This suggests that minimally structured hypertexts are better for specific learning tasks than for gaining overviews of new topics. The structured hypertext was easier to use and produced more exploration, suggesting that content map design influences whether readers take advantage of hypertext's flexibility.
A Hypertext Interface to Relational Databases BIBA 8
  Martin Durr; Stefan M. Lang
In the past, database systems (DBMS) have served as a powerful tool to support a variety of tasks with a high degree of concurrent access to shared, persistent data. The large number of advantages has led to their introduction in application areas with extensive requirements -- such as cooperative environments like CASE, CAD or computer-aided instruction. Here, however, the priorities of the features are shifting towards casual user support and browsing support. In the poster, we propose an approach that adds hypertext features to existing relational databases. The main idea is to establish an interpretation component on the relational database that interprets the tuples and tables in an hypertext manner, allowing the user to work with them as if they were nodes and links. As a side effect, navigation in the hypertext fashion becomes possible on the relational data. The data itself remains unaffected. That is, it is not changed in any way by the interpretation component. Other users can access the relational data in the original way (i.e., by using SQL). Finally, the approach is generic. That is, it does not rely on any particular database, but it is applicable to any relational repository. The interlinking services are established by a rigorous analysis of the database system's data dictionary.
Extension of Hypertext to a Conceptual World: Concept Browser for a Personal Information Base BIBA 9
  Hiromichi Fujisawa; Hidefumi Kondo
An extended form of hypertext which has a "concept network" as a knowledge base is proposed. While the conventional hypertext system has association links only between units of media information, a knowledge-based hypertext system we propose here can store "concepts", "relations", and facts that are represented in terms of those concepts and relations, and links of this system also are bound between concept nodes and media information. Concepts form a taxonomical hierarchy of things which are pertinent to information to be stored in the system. Some nodes represents media information. Relations are predicates that can represent attributes of things, and facts such as "Mr. A received Prize B in 1990." A prototype, ConceptBrowser, has been developed to apply it to a "personal information base". The system supports registering new information, browsing the contents of the knowledge base, and retrieving concepts and media information. One of the attractive features is associative retrieval of concepts. We believe this kind of system can amplify information processing power of creative knowledge workers such as designers, engineers, researchers, writers, etc.
Joining Ideas BIBA 10
  Geri Gay; Deborah Trumbull; Joan Mazur
This poster reports on exploratory research that examines how students used a hypermedia program. Outcome measures include how students learned to use the system, the strategies they used to search for information, the amount of relevant information located by their searches, how information was used in their final essays, and user perceptions about the system and their use of it. Forty-one students used four search modes (Browse, Index, Guide, or a Mixture) for searches of a highly visual, interactive program. Findings suggest that designers develop a variety of interfaces to facilitate user searches, while attending to user needs, task and the environment. Crucial questions are raised about the use and interpretation of visuals for content representation and as organizers (metaphors).
Dynamically Created Guided Tours in Hypertext for Learning BIBA 11
  Catherine Guinan; Alan F. Smeaton
The aim of our work is to devise methods for searching through hypertext and presenting the nodes in a logical manner. We believe that implementing a dynamic guided tour facility which would automatically choose the best route for a user to take based on his or her search query is the next step after static guided tours and the use of information retrieval techniques to find start nodes for a user. The dynamic tour has the advantage of being directly related to the needs of the user where only relevant nodes are displayed as opposed to the static case where the author decides what is relevant before the users have even formulated their queries. As our application is to be used in a learning environment, the order of presentation in the hypertext is very important. There is no point in selecting the most relevant nodes and displaying them in random order if the user is trying to learn about the subject matter in a constructive fashion. To this end, we have introduced 'link types' to our system. There are ten distinct link types, some of which are, 'is_a', 'consists_of', 'precedes' and 'facilitates'. Using these types we rearranged the presentation order so that a logical sequence of nodes was displayed to the user. There should be a natural progression from the most basic information to the more complex so that the user does not become confused with the material.
Applying Cognitive Apprenticeship to the Design of a Hypermedia Learning Environment: The Lab Design Project BIBA 12
  Peter C. Honebein
The Lab Design Project (LDP) is a hypermedia learning environment created with Claris's HyperCard 2.0 running on Macintosh computers. The hypermedia environment is comprised of 130 megabytes of architectural blueprints, color photographs, transcribed interviews and scanned documents which create a simulated biotechnology building. This environment enables students to practice sociological research from the perspective of a sociological researcher investigating how building design influences the people who work within the building.
   The task of the students in this environment is to develop a sociological research question through the exploration of the hypermedia environment (in essence, they "walk through" the building, using a blueprint navigational system to go into labs, view work benches, look at equipment, talk with scientists in their offices, etc.), then answer that question by linking information together. For example, a student might question the existence of desk lamps in a lab. The student would search for answers to his question by reviewing documents and interviews to determine why desk lamps are in the lab. When the student finds an answer, the student links the supporting information to the picture of the lab which illustrates his question. Since the environment is collaborative, other students can access the link, comment on it, and build interpretations of the building design. The students work is continually supported and evaluated by the professor and research assistants.
   The developers of the LDP hypermedia system will be present during the poster session to discuss the pedagogical aspects of the system, namely the prescriptions of cognitive apprenticeship and their influence on hypermedia design for education.
A Measure of Hypertext Linearity BIBA 13
  Mark A. Horney
This study of 8 hypertext authors uncovered a conflict between the two common ideas: (a) that hypertext is non-linear; and (b) that hypertext readers are free to choose their own reading sequence. Trail records showed some hypertext users acting in regular, linear patterns even within large associative webs, and others working nonlinearity in sparse documents. These actions were quantified by a metric determining average ancestral path lengths for nodes as visited by particular individuals. This result suggests the concept of non-linearity must be applied separately to hypertext documents and to the hypertext reader/authors using those documents.
WITH -- A Project on Computer Supported Hypertext Construction BIB 14
  R. Kuhlen; R. Hammwohner
Hucklefine Defined BIBA 15
  Mike Mosher
...discussed in light of explorations of randomness as an organizing (or disorganizing) principle in various arts a literature.
Modeling Distributed Hyperdocuments with Markov Chains BIBA 16
  Charles Nicholas; A. Brooke Stephens; Yelena Yesha; Keith Humenik
Suppose that the nodes in a distributed hyperdocument are scattered over some file servers. Markov Chains can be used to model the performance of such hypertext systems, but calculation using existing techniques is impractical for hyperdocuments with more than about 1000 nodes. We introduce heuristics, which can be calculated in O(n) time, for estimating the relative frequency with which nodes are visited. Then, the most popular nodes can be placed at centrally-located sites, reducing communication costs. In our simulations, these heuristics are consistently within one percent of the value produced by the O(n²) Markov Chain techniques.
Using a Hypertext-Based Index to Access Hardcopy Documentation BIBA 17
  Duane Ressler
This poster illustrates how a hypertext-based index was used to provide integrated access to information in a hardcopy documentation set and to provide a structure for gradually moving portions of that documentation set online. The poster also shows the advantages that a hypertext interface can provide to the user of an online index, particularly when the index is always available as the user works with an integrated system of software products. In this way, the index becomes the support structure for a large documentation library, providing quick access to information as an online front end to hardcopy documentation.
The Hyperbase Developer's Toolkit (HDT) BIBA 18
  John Robertson; Kai Foong
The HDT has been developed to assist developers of hypersystems convert linearly structured text into a form suitable for hypermedia databases. This transformation process is classically one of the most expensive operations in hypertext system production.
   The tool provides a shell which guides the human editor through this development. First, by providing procedural direction and second, by assisting with the identification of the hypertext components within the linearly structure document.
   By providing software tools to assist the human in this editorial process, we believe that conversion costs can be reduced, thus improving hypertext systems commercial viability.
   The system also provides us with a platform to explore different conversion methods. From this work we hope to identify better procedures for producing hypertext databases from paper-based text.
Learning to Read a Hypertext: A Cognitive Approach BIBA 19
  Jean-Francois Rouet
This study examined the learning and use of a simple hypertext system by secondary school students. In the course of two training sessions, sixty subjects aged 11 to 15 were asked to answer several series of questions varying in explicitness and complexity, by searching a hypertext. We observed a significant increase in the quality of answers / search time ratio. Furthermore, subjects tended to devote a greater proportion of search time to the selection process, but only for implicit or complex questions. It is concluded that, in order to use a hypertext efficiently, inexperienced readers have to build up specific reading strategies.
Hypermedia as a Training Tool for Echocardiogaphy: An Empirical Evaluation BIBA 20
  Marc M. Sebrechts; Kathryn Permenter
The utility of hypermedia for training in echocardiography was evaluated using three learning tools. A "linear" system provided a fixed sequence of cards that controlled the presentation of videodisc-based case studies. A "structured" system provided access to the same images via an outline, organized by view and diagnostic entity. An "enhanced" version added alternative navigational strategies plus schematics, textual notes, echocardiograms and associated stethoscopic sounds. A group of fifteen cardiologists showed a strong tendency to restructure the information in a familiar linear fashion. Image recognition and diagnosis was roughly comparable across training groups, although the enhanced version did show an advantage on a test of related conceptual knowledge.
Finnegan's HyperWake -- The Clicky Way to Waking BIB 21
  Andrea Ventura
Versioning Issues in Hypermedia Publishing Environments BIBA 22
  Anja Weber
Effective support is required for the maintenance of the final and interim data that arise incrementally in cooperative publishing. Publishing can be characterized as an open-ended design-task. As a consequence, the design products, i.e. the produced hyperdocuments and especially their interim states, cannot be described sufficiently declaratively by attributes. The question arises, how the partners involved in the publishing process will find those versions of hyperdocuments, that fulfill their current information needs.
   Our approach to this problem is to maintain contextual information with the versions of hyperdocuments. We propose a hypermedia version server that differs from other approaches by
  • (1) maintaining the derivation-history of hyperdocuments consisting of nodes,
        links and composites across document boundaries,
  • (2) offering a task concept that allows coordinated, task-oriented change
        management with hyperdocuments, and
  • (3) keeping the influence of comments on the creation of new versions and the
        set up of new tasks. All this kind of recorded information can be employed to identify versions of hyperdocuments and thus improves access to versions of hyperdocuments.