Virtual library - paths to knowledge

Abstract

The paper focuses on the use of elaborated computer visualization techniques to support multi-user browsing within a digital collection of library books and subject categories. First, a short description of the research project is given, and the two different interfaces of the developed application, called `Explore Interface' and `Browse Interface' are described. Then, some of the theoretical issues and practical use of the interfaces are presented. Finally, the experiences gathered from this research project and their impact on possible future working directions are summarized

Full text

Page 1

Virtual Library - Paths to Knowledge


Malgorzata Bugajska, Ph.D. Student [bugajska@arch.ethz.ch]
Andrew Vande Moere, Ph.D. Student [moere@arch.ethz.ch]
Chair for Architecture and CAAD, ETH Zurich
HPI F 22 ETH Hönggerberg , 8093 Zürich Switzerland
Tel: +41 1 633 29 97


Abstract
This paper focuses on the use of elaborated computer
visualization techniques to support multi-user browsing
within a digital collection of library books and subject
categories. First, a short description about the research
project is given, and the two different interfaces of the
developed application, called ‘Explore Interface’ and
‘Browse Interface’ are described. Then, some of the
theoretical issues and practical use of the interfaces are
presented. Finally, the experiences gathered from this
research project and their impact on possible future
working directions are summarized.


1. Introduction

The phenomenon of cyberspace still inspires research
projects in science, art, business, and architecture. For
Michael Benedikt (1991) the so-called ‘cyberspace
program’ should start experimentally, by creating ‘crude’
and ‘fragile’ cyberspaces with a limited number of users,
from which the most essential lessons can be learned. He
foresees this process taking decades of time, meanwhile
stimulating spin-offs in various areas of computing and
engaging research people with different kinds of
experiences and education. All with one goal: to make the
visionary cyberspace a fact of reality, “Because the design,
institution, and management of cyberspace will be a task
of immense scale and complexity, it can be simply be
argued that it is never to soon to begin” [1].
In the spirit of Benedikt's thoughts, this paper focuses on
the use of elaborated computer visualization techniques to
support the process of multi-user browsing within a digital
collection of library books and subject categories. The
concept of researching a certain area of interest, by
browsing through a collection of knowledge-based, spatial
paths within a three-dimensional shared information space,
was the starting point of this research project.
Furthermore, the intrinsic qualities of two different digital
user-interfaces were explored. Specifically, the proposed
application should enrich the use of a primarily
hierarchical cataloging system and simultaneously give the
users additional opportunities to discover and share
knowledge with others in an intuitive way. With the
awareness that efficient
interdisciplinary information retrieval play an increasingly
important role, the “Virtual Library - Paths to Knowledge”
project proposes an experimental way to trace, visually
represent, and share our global knowledge in a unique
way. The presented application was created as a part of the
exhibition for the 'Library of the Future' event, held at the
ETH Library between December 1999 and March 2000.

2. Project Description

2.1. Considerations

The means of sharing information and knowledge between
the users of the digital library as well as the particular way
users interact with a strict classification system became
important starting points for developing the interfaces for a
data-intensive virtual library system.
We started our research by observing how daily users
interact with physical collections of documents and by
focusing on the implications of physical document
location on the process of perceptive information
browsing. We concluded that sophisticated, decimal
classification systems are too abstract to help guide
common library visitors in browsing through the huge
collection of subjects. When access to the physical
collection is permitted, the easiest way to locate a group of
interesting documents is by indicating the physical area
where this particular range of collection is located. While
interacting with the collection of documents, a patron
(physical library user) may examine the covers, browse
through the titles (this task is determined by the size and
shape of the lettering), or become interested in a document
with a familiar publisher's logo, even if it does not belong
data exchange and

Page 2

to the area of the initial interest. Notably, this procedure
enables patrons to become easily inspired by other
documents in the immediate proximity of the initially
searched book. Important factors in this decision making
process are the typical ordering by subjects, which is
reflected in the physical locations of the books, and their
unique physical appearance. For example, heavily used
and worn out documents (such as hard copy documents)
can be particularly intriguing,
distinguishable physical look can indicate a certain
seminal value of the document. Books that include many
user annotations can be of an interest too, demonstrating
the potential of common interests among the users of the
library.

2.2. Aim

As previous work in the area of information visualization
has shown [5, 9, 11, 13], the structural use of dynamic
imagery and animated events enriches the human ability
for interconnecting experiences, organizing information,
and sharing the current state of the information with other
users.
The visualization system called Vineta [6], developed by
the University of Nottingham, demonstrates the capability
to represent a large number of dimensions, retrieved from
a certain collection of documents, by applying
sophisticated visualization metaphors.
In the specific area of user interfaces design for digital
libraries, the importance of the concept of ‘direct
manipulation’ was shown in the concrete example of the
Visible Human Project [7].
A further interesting project that deals with the
visualization of a collection of interconnected books is
called VR-VIBE [2]. This fully three-dimensional
environment enables users, represented as animated
avatars, to navigate freely through the presented data
structure while selecting documents, performing queries,
applying certain filters
information.
The aim of our Virtual Library project however, was to
concentrate on the area of creating a digital counterpart for
the physical and spatial qualities of the library, and to
enrich a collective exploration of the documents in an
online digital collection. Furthermore, we used the concept
of three-dimensional information mapping in an attempt to
communicate informational values towards the user in a
cognitive way. Finally, it should be remarked that these
interfaces were especially designed and carefully
developed to give users an opportunity to make unusual
and inspiring choices by allowing them to operate almost
exclusively on the visual level.



as their fairly
or requesting additional
2.3. Application

The Virtual Library application envisions an alternative
way to visualize and explore a digital library catalog. This
project, experimental in nature, demonstrates the rendering
of the mathematically ordered data into meaningful spatial
visualizations by ‘mapping’ that data onto a three-
dimensional virtual space. The abstract interpretation
concept being used is based on a numerical classification
standard of the refined hierarchies among subjects. This
representation mechanism makes it possible to transform
parts of the large library database into a system of
comprehensible values, so
unconventional exploration metaphors can be applied.
Practically, the concept of 'knowledge paths' was applied
upon a small, re-engineered catalogued collection of
newly-arrived books (about 3000 entries), which we
retrieved from the ETH Library. By applying different
filtering algorithms, we were able to build a meaningful
database of interrelated books by using only the book titles
and their corresponding decimal system numbers. It should
be noted that the use of keywords was completely
neglected, and were therefore not available for the user as
means of querying the gathered data. During the library
exhibition event, two distinctive interfaces were designed
and displayed on two separate computers, positioned in
opposite corners of the provided area.

that interesting and


Figure 1. LINE, Library for the Information Era,
visualizations of the knowledge paths generated
by the users of the library. (outside and inside
view)

Page 3

Notably, the decision for creating two separate interfaces
was not merely a result of the physical exhibition layout
but represented an additional conceptual part of the project
itself.

3. Application

3.1. Paths of Knowledge

The process of revising and classifying our gathered
knowledge plays an important role for many reasons.
Essentially, it gives us an opportunity to have a global
look at our heritage and discover new interesting areas for
future inventions.
In the Information Era of today, in which the global
amount of data grows exponentially and the processing
hardware becomes continuously more available, this
process occurs even faster and with much broader scope.
New disciplines have already been created, researched,
and documented, based on some completely new ideas and
needs generated from this collection of meanings. In this
sense, tracing and visualizing user activities and data
processing within data-intensive catalogue systems could
be a significant source of meaningful information [9]. For
this goal, the concept of ‘Paths of Knowledge’, based upon
a research project called LINE (Library for the
Information Era) [4] is proposed as a new kind of a three-
dimensional data-carrying visualization metaphor that
would allow tracing, sharing and enriching the user
activities within a certain data collection. Ultimately, these
developed geometries grow on the edges of currently
defined disciplines, focusing on new interdisciplinary
research areas that have already become crucial for
scientific progress.
Conceptually, Paths of Knowledge are collectors that
accumulate products and methods of investigation. It
would be interesting to see these methods (or patterns of
their usage) being traced in the future to formulate unique
ways of organizing information in general. Such defined
paths can also be seen as continuous spatial traces that
record the latest moves in interdisciplinary science and art
fields, while allowing for the discovery of possible
connections among them. It should be noted though, that
since they record vast amounts of individual investigation
results, their effectiveness intrinsically depends on the
skills of their creators. In this project, we consider these
Paths of Knowledge to be spatial metaphors that visualize
the activities occurring inside the applied library
classification system or
Conceptually, these ‘paths’ represent the accumulated
experience of library users, enabling users to ‘visit’ the
data-worlds of others. By using such conceptual traces,
users collectively build an open-knowledge base for
further development, sharing the information about certain
subject areas and forming methods or patterns of research.
‘boundary object’ [3].
3. Explore Interface

3.1. Personalizing the Classification System

One of the most efficient ways to browse within an
electronic library catalogue is to select a predefined,
hierarchical thematic category like ‘Arts’, ‘Arts-
Architecture’, ‘Art -> Architecture -> History’ or a set of
keywords describing the desired subject (architecture,
history, …). The library classification system1, as any
instance of classification, is created to accommodate many
communities of practice and satisfy the informational
requirements of each of them [3]. We decided to challenge
the power of the classification system and let the users of
our experimental application fracture the existing scheme,
letting them create their personal classification categories
while looking for documents with interdisciplinary
character. As a result of the interdisciplinary character of
scientific literature available in the ETH university library,
a considerable number of books in the database possessed
a multiple category assignment. This means, for instance,
that a certain book about famous churches in Rome can be
a simultaneous member
hierarchies, such as ‘Architecture-History’, ‘Construction
Engineering-Conservations
‘Tourism-Travel Guides’ or ‘National Italian Misfortunes-
Catastrophes’. It should be noted that this specific
phenomenon triggered the chosen design of the Explore
Interface, as it enables the users to specify their personal
interrelated subject collections.

3.2. Visualization of textual information

The applied classification system is presented as a
hierarchical list of disciplines or categories presented in a
keyword-like language. Some of the categories are very
descriptive and represent almost a long sentence.
Navigation and user manipulation with such interface
elements requires a significant amount of time, as textual
descriptions need longer attention time to decode and
understand than, for instance, iconic representations. Since
the text has to be decoded through reading, it needs to be
represented clearly on the screen and displayed by
contrasting it with the overall background, as well as have
a proper size [12]. Presentation of the hierarchical lists
involves an aesthetical and ecological approach, which
of several classification
of Historic Buildings’,

1 "The Universal Decimal Classification (UDC) is an international
scheme which endeavors to cover all areas of knowledge. Its origins lie in
the Dewey Decimal Classification (DDC) which was adapted towards the
end of the 19th century by Paul Otlet and Henri LaFontaine in an attempt
to create a universal bibliography. […] The scheme consists of 60,000
classes (divisions and sub divisions) as well as a number of auxiliary
tables to describe countries, etc." Link was visited on Nov. 2000.
http://www.ukoln.ac.uk/metadata/desire/classification/class_3.htm

Page 4

means that visual elements have to be used scarcely but
effectively, to give the user a sufficient level of interface
functionality and comprehension. This point of view is
supported by James Wise (et. al.) who says that: “The
bottle neck in the human processing and understanding of
information in large amounts of text can be overcome if
the text is spatialized in a manner that takes advantage of
common powers of perception.”[16].
One of the issues in using text as an active element within
the interface involves deployed rendering technology.
Practically, the visual interface is built using the VRML2.0
(Virtual Reality Modeling Language) language. It can be
noted that the VRML-world text displayed on a computer
screen does not appear to be smooth or anti-aliased.
Consequently, there is a certain distinguishable critical
value of size and color properties of text elements that
makes them easily readable.
Another interesting characteristic of this interface is the
considerable amount of screen space that users are
provided with for manipulating the elements. It was
chosen not to construct a display that appears to be
crowded with textual information, as this would have
implicated a lack of organizational data structure and
therefore would have adversely affected the user-
friendliness of the interface. Instead, the different
classification hierarchies were structured as lists and
presented as animated pull-down menus. These specially
designed animation events reveal the sub-categories, and
at the same time emphasize the idea of two conceptual
axes of the interface.

3.3 Animation events

Animation is a powerful, functional component for data
intensive user interfaces. By shifting the user's task from
cognitive to perceptual activity, it frees up his cognitive
processing capacity for application tasks [8]. Assigning
different animation events to the same object of the
interface helps to communicate to the user the state of the
activity he is involved in [5].
Animation also helps us to convey the illusion of a
temporal-spatial notation for the history of the user's
interaction with the interface. It is especially helpful in the
case of interfaces, which work with qualities of both two-
dimensional and three-dimensional worlds. Both interfaces
presented in this paper have carefully designed animation
event sets that support the user-friendly interaction with
the information appearing on the screen. The ‘animation
set’ of the Explore Interface consists of three independent
‘animation events’. These events work on directions drawn
by Cartesians axes. Within the plane of the screen, a
vertical axis represents a direction of the animation event,
which reveals the sub-categories and is a part of the pull-
down menu function. The ‘depth’ axis, located
perpendicular to the plane of the screen, has been chosen


Figure 2. View on the Explore Interface after the
user has proceeded with a category selection.

to direct the event that communicates the history of the
user choices while browsing.
Two animation events have been assigned to the selection
frame. This is an outcome of a functional characteristic of
the selection frame, which serves as an element informing
the user about both his future and already performed
actions. The decision to proceed with the selection of
collected disciplines, or going ‘forward’ triggers the
appearance of a new selection frame.
The animation associated with this action is designed to
minimize the effect of a sudden disappearance of one
element and appearance of a new one on the interface. The
old frame disappears by giving the user an illusion of
‘stepping through’ it. In the meantime, the new frame
evolves from a point to a horizontal line to a rectangular
selection frame. The vertical axis serves as the rotation
axis for the selection frame.
The frame rotates when the user decides to take one step
back, to revise or refine his previous actions. The selection
frame rotates 180 degrees, demonstrating the spatial
quality of the current procedure. This activity conceptually
'makes' the space for the elements from the previous
selections as they reappear. As a result, they co-exist with
the elements from the current selection and are still
slightly visible in the background of the screen.

Page 5

Figure 3. View of the Explore Interface with a
dynamic horizontal rotation of the selection
frame after the user chooses to go back in the
history of decisions.

3.4 Interface Description

The Explore Interface offers the possibility to browse
through the different top- and sub-categories of the library
collection in a simple, text-based view. The spatial
position of categories within interface is generated
randomly. Animated pull-down menus enable the user to
explore or select directly related subcategories of a certain
subject field, simultaneously avoiding unnecessary text
clustering. Inside this open interface screen, users are able
to select multiple categories of their choice by simply
clicking, dragging and dropping
connotations inside a central selection frame. A rapid and
easily distinguishable color change of the chosen text tags
gives the user immediate feedback of his actions. After
this selection process is over, and thus two or more
categories are chosen, the interface makes it possible to
view the ‘deeper’ categories by clicking on the ‘+’ or ‘-‘
symbol on the menu bar. After such a user decision, the
program starts to navigate through the previous user-
selection and displays a more detailed data view of the
chosen query. Subsequently, all selected themes become
more specific and a small set of relevant books enters the
newly presented view. Now the user is able to perform the
the appropriate
same actions again but this time with an option to select
the just appeared books. This procedure can be repeated in
both conceptual directions to narrow or widen the
categorization level or change the number of chosen items.
It should be noted that if the plus sign is pressed, the
interface ‘flies’ almost literally forwards, turning the depth
(the axis perpendicular to the screen) into a metaphorical,
virtual time axis. Conceptually, this action will shift the
interface one level lower in the hierarchical library
database structure, to search for the related decimal sub-
categories.
At the same time, relevant books of the categories that
were chosen are also searched by the system and enter the
interface for further user evaluation. Subsequently,
clicking the minus sign will retract the interface backwards
on the time-axis, enabling the user to see a full-view of the
previous menu choices while simultaneously showing the
last interface in the blurred background. The intuitive and
direct notion of this time axis proved to be useful to orient
the user in the personalized decision process.



Figure 4. Functional features within the Explore
Interface.