The Pausanian Notation: 
a method for representing the structure and the content of 
a hyperdocument 
Nickolas Gouraros and Wendy Hall 
Department of Electronics and Computer Science, 
University of Southampton Email: ng94r@ecs.soton.ae.uk, wh@ecs.soton.ae.uk 
ABSTRACT 
This paper introduces a new approach for improving 
hypermedia design, by providing the author with a tool to 
visualise, examine, and analyse the structure of documents 
containing hypermedia links. Our proposal is a new 
representation method the purpose of which is not to show 
the document structure in graphical form in order to enable 
the users to know where they are, where they can go next, 
or to give an overview of their environlncnt \[12\]. The 
purpose of our representation method is to depict what 
options the users are about to be offered, so that the author 
can examine the structure in order to better guide theln 
around the information space. Our representation is shown 
in the tbl'm of a map consisting of a proposed classification 
of elements that define the composition of a 
hyperdocument. To extract and classify those attributes we 
have analysed in detail all three dimensions of the 
hypermcdia cube: internal dynamics, external visual 
appearance and content synthesis. 
After introducing our diagrammatic notation we give a brief 
example of how it could be applied to an existing 
hyperlnedia application. In the boundaries of this paper, we 
include the part of the method which represents the 
structure of just one hypermedia document and its linked 
branches. The paper concludes with. a brief description of 
how the lnethod is applied to represent groups of 
hyperlnedia documents, and a discussion of our plans for 
future work. 
KEYWORDS: Hypermedia design, authoring tools, 
representation methods, hyperlnedia maps. navigation, 
hypermedia writing. 
1.0 INTRODUCTION 
Models or methodologies such as OOHDM \[30\], RMM 
\[17\] and HDM, \[10\] have been applied very successfully in 
the process of hypermedia application design by expressing 
in a non-ambiguous way \[28\] the relationships among the 
interconnected objects. Their success solved many design 
problems and proved formally the need of structuring the 
design process. 
Two-dimensional, three-dilnensional, n-dimensional, 
hierarchical and non-hierarchical structure representations 
of hyperlnedia networks, have been the subject of much 
research, as well as commercial products for many years. 
Generally. their main goal has been to provide 'the user' 
with a tool which will solve problelns such as disorientation 
and being lost in hyperspace. 
However, the questions we need to ask but which have not 
yet been suitably answered are: "what causes 
disorientation? why are users allowed to becolne lost in the 
hyperspace; why do we still, and how long we will be 
designing applications witll navigational errors?" The 
hypermedia field has produced some excellent 
representations -in the forms of overview diagrams or 
maps- but, 'to tell a multitude of little white lies; they 
suppress the truth to help the user see what needs to be 
seen' \[23\]. Most of those maps have been "system- 
generated means of reader orientation' \[20\]. 
2.0 THE PAUSANIAN NOTATION APPROACH 
The need for another approach to the representation of 
hyperlnedia networks derived from our research and the 
development of the Pausanias 2000. a hypermedia library 
for the ancient Hellenic World (The Attica Region) based 
on the Microcosm hypermedia systeln \[81. The application 
at its prototype stage, includes 140 content categories, 
including links between 300 text documents. 2000 pictures, 
250 region maps, 3D graphics in a VR environlnent, video 
and sound sequences etc. As authors, during 
ilnplementation, we felt that we needed a tool which would 
allow us. to keep track of, and control the entire structure. 
A way to achieve this. is obviously to be able to visualise 
the hyperdocuments. Having exalnined CUl'rcnt 
representation inethods \[5,7.9, I 1.12,13.18.22,24,25, 
27,29,32\] we realised that the majority, if not all of them. 
are implemented as an aid to users rather than the authors. 
By combining the philosophics of a design model with an 
overview diagram we began to develop a representation 
method, that is a tool lbr helping the author, to improve the 
overall structure of the hyperlnedia application, and to 
avoid inconsistencies and navigational crrors that confuse 
the user. The Pausanian method is a diagrammatic notation 
for representing the static and dynamic properties of a 
single or a collection of hypermedia documents in order to 
help the author to visualise the structure they are creating. It 
employs a wide range of shapes and symbols based on the 
characteristics that make up the structure of a document 
containing hypermedia links (figures la.lb). With a clear 
diagrammatic representation and an overall picture of the 
different issues concerned with the design of the 
hypermedia application, the author can save considerable 
amount of design time by evaluating or examining the 
prototype in order to minimise the amount of serious 
• navigational and interaction errors that often occur \[26\]i 
The name Pausanian derives from Pausanias, a Greek 
historian, traveller, and geographer. During the 2'" century 
AD he was travelling around Greece and describing all the 
sites and monuments he was seeing. His work was called 
the "Description of Greece" which gives a detailed 
account of the monuments of art and of the legends 
cotznected with them. 
Tile Pausanian notation is divided into two main sections. 
The first section represents one hypermedia document, and 
the second one represents a group of selected hypermedia 
docurnents. To visualise the structure of one hypermedia 
document we distinguish those elements that, we believe. 
characterise it (aspects of interest). These elements are 
drawn diagrammatically in what we call the "Pausanian 
Map fo," representing the structure of a single hypermedia 
document". To visualise the structure of a group of 
hypemedia documents, we define a set of variables 
(variables of interest) which are statistically represented in 
what we call the "Pausanian Charts for representing the 
structure of a group of hypernledia documents". 
3.0 ASPECTS OF INTEREST 
We call the characteristic elements that define the structure 
of a hypermedia document and are appropriate to the 
hypermedia author's interest, the aspects of i~uerest. The 
aspects of interest within a hyperdocument structure are 
classified into three main categories: The 'endogenous', 
which include aspects that characterise the internal 
structure or dynamics of the hypermedia document, tile 
"exogenous' which contain aspects that define its visual 
appearance and the 'auctorial" category which is concerned 
with aspects of interest on tile content's structure and 
composition. All aspects of interest are assigned with 
attributes graphically represented in the Pausanian map. 
Authors are provided with the pre-defined attributes, but 
there are cases where they are also given the opportunity to 
include their own. 
3.1 ENDOGENOUS ASPECTS OF INTEREST 
Tile representation of the endogenous aspects of interes{, on 
the Pausanian map. produces an 'x-ray' that depicts the 
internal structure and dynamics of a hypermedia document. 
By examining the map, the author can clearly perceive all 
objects that comprise the structure as well as the dynamics 
effecting it. At the top level we classify the endogenous 
aspects of interest as: types of nodes, types of links, types 
of anchors, location of anchors, status of traversal, status of 
direction and depth of path. 
Each endogenous aspect of interest is characterised with a 
set of attributes acting as identification marks on the 
Pausanian Map. The classification of attributes for the 
en&~genous aspects ~f interest is as follows: 
3.1.1 Type of Node 
Tile 'type of node" justifies its role as an aspect of interest 
to the hypermedia author, because it defines the identity of 
all the different objects that comprise the structure of a 
hypermedia application i.e. text, picture, sound etc. Each 
type of node together with its attributes will be represented 
in the Pausanian map so that tile author will be able to 
clearly distinguish all information units and the way they 
are presented to the user. Since the Pausanian notation is 
currently implemented on Microcosm based hypermedia 
applications, eight categories have been assigned, as the 
types of nodes that can be currently viewed in Microcosm: 
Text, Graphic. Audio, Video. Animation, Graphic Menu. 
HTML( any HTML document, viewed by the Microcosm's 
Web viewer), VR (any virtual reality objects) viewed by the 
Microcosm's 3D viewer, External Object (any independent 
application linked with Microcosm). Since each type of 
node has different characteristics (in terms of comprising 
the structure of a hyperdocument), it has been assigned with 
attributes which identify it as an aspect of interest to the 
hypcrmedia author. 
Type of Node: TEXT 
Type of TEXT: includes Core Docmnent, Section. 
Supplement, Question - Answer (attrb - Interactive. 
Passive), Table (attrb - Dimensions). List-menu, Footnote. 
Reference, Glossary. Dictionary, Bibliography. 
Combination of (Type of Text + Type of Tcxt), other by tile 
author. 
Type of Node: GRAPHIC 
The type of node graphic is given two sets of attributes, one 
that identifies what type of graphic it is and another one 
that defines the semantic relation between the graphic and 
its source, when tile graphic is a picture and serves as a 
destination node. 
Type of Graphic: includes Picture. Diagram, Graph-Chart. 
Graphic Combination. 
Status of Semantic Relation when tile graphic serves as a 
destination node includes: Explanatory, Decorative. 
Allegorical I2\]. 
Each of tile types of graphic attributes arc given an 
additional identification mark which represents on tile 
Pausanian Map the case of any text accompanying the 
graphic. 
The type of node: AUDIO 
The attributes that define an audio node and are appropriate 
to the hypennedia author's interest include: tile type of 
audio, which reveals the content of the particular audio file. 
For example, Verbal description, Music. Composite: the 
length of the sequence which represents the total playing 
time of the audio file; and the status of the audio player's 
(viewer) controls, which show whether they are available oz" 
not to tile user while they are listening to audio information. 
The type of node: VIDEO, ANIMATION 
The attributes that define kinetic information \[19\] such as 
video and animation and are appropriate to tile author's 
interest include: the length of the sequence which 
represents the total playing time of a video or animation 
information unit; the status of the video and animation 
player's (viewer) controls, which show whether they are 
available or not to the user while viewing the particular 
video or animation; and the status of interaction, which 
represents whether a sequence is interactive or passive. 
The t37~e of node External Object is assigned two attributes: 
the type of External Objcct which states its identity e.g. 
'database', and the status of interaction representing 
whether that particular external object allows user 
interaction or appears passive. Attributes for the types of 
node Graphic Form- Menu, HTML and VR are under 
research. 
3,1.2 Types of Links 
One of the most important elements in the structure of a 
hypermedia document, is the different types of links 
provided by the hypermedia system. Microcosm includes, 
four associative types of links, one visible (button link - 
highlighted anchor) and three invisible (specific, local and 
generic - their anchors are not highlighted). At this point we 
do not consider other dynamically generated types of links 
(i.e. Microcosm's compute links). 
Button 
A link fl'om a visible anchor in a source node to a specific 
location in a target document. It is used in situations where 
the link should be obvious and the author wishes to indicate 
visibly its prescncc; for example a link from the anchor the 
first Athenian Coin, to the associated picture of the actual 
coin. 
Specific 
A link from an invisible anchor in a source node to a 
specific location on the target document. It is used in 
situations where the link provides detailed information 
about a particular subject "not strictly related' to the overall 
content of "the source node" but directly related to its 
anchor. Authors would use the specific link when they wish 
to offer the particular information but also allow the users 
to study the content without interrupting their cohesion; for 
example a link fi'om a reference: Aristotle's description of 
King Archon's role, to the actual document is rclErred to 
as Aristotle's work "Athenaion Politeia" line 57. The user 
has already read the text block in the source node where the 
author discusses the King in ancient Athens and if they 
wish to examine the actual reference they can trace a 
question (select the text block and ask Micrososm to 'show' 
the link connected to that phrase) and follow the link to the 
Aristotle's work. 
Local 
A link from an invisible anchor(s) in a source document to 
a specific location in a target document. When a phrase or 
word has been linked as a local link, all occurrences of this 
particular anchor in the source node are linked to the same 
destination node. It is used to ensure that users can only 
reach certain documents by careful research, or to avoid 
presenting thern with inappropriate intbrmation. Suppose 
the word harbour in a text node about the ancient harbour 
of Piraeus, is selected as the source anchor and a link is 
made to a picture painting of the ancient harbour. If the link 
type local is selected when the link is made 'all" 
occurrences of the word harbour "in the source document 
only" are linked to that painting. 
Generic 
A link from an invisible anchor to a specific location in a 
target document. When an anchored piece of text has been 
linked as a generic link, all occurrences of this anchor in 
any document throughout the application, are linked to the 
same destination node. It is used in situations where the 
link should be available 'throughout the application': for 
example a link from the word 'archon" to its explanation in 
the ancient Greek-English dictionary. 
Sequential 
Another type of link included in the Pausanian notation is 
the sequential link. It is represented on the Pausanian map 
when a visible anchor (anchored text, anchored graphic 
object, or anchored graphic selection) leads the user to a 
single linear path of related nodes. It is therefore a 
navigational rather than an associative link. The hypertext 
author uses such links when he wishes to guide readers into 
a logical path in order to achieve a specific objective. A 
source node can include visible and invisible links as well 
as a sequential link(s) defining guided tours specifically 
created by the author. Sequential links provide a more 
limited version of what is available within the structure of 
the particular node \[15\]. That linear-logical path consists of 
nodes including a tEw or no button, specific or local links, 
only anchors indicating 'go to Next' and 'go to Previous" 
nodes. In the case when there is a sequential link starting a 
linear path fi'om a graphic type of node. then only the 
maximum depth of path will be represented on the map. 
(for the aspect of interest depth of path see later). 
Three other attributes attached to each type of link are the 
Status of Travcrsal. Status of Direction and Target at the 
Destination document: 
3.1.3 Status of Traversal 
One to One 
The source anchor leads the user to only one destination 
node. 
One to Many (Indirect) 
The source anchor leads the user to a choice of destination 
nodes, e.g. the source node is about the ancient Greek 
philosophers and in the tburth paragraph the name Plato is 
a visible anchor linked to a choice of three different 
destinations: a document about his life. a picture of a 
statue and another document about his works. 
One to Many (Direct) 
By following the source anchor the user is lead directly to 
more than one destination nodes, e.g. the source node is 
about Daily life in Athens at the 5"' century BC; a visible 
source anchor is the words women's embellishment. If the 
user follows the link, three nodes will appear: a picture of a 
vase painting showing such scenes as friends preparing a 
bride for her wedding, a picture of a vase painting showing 
women applying perfume to themselves, and the third, a 
text node discussing embellishment in ancient Athenian 
Society. 
3.1.4 Status of Direction 
One way Link 
The source anchor leads the user to a destination node from 
where they cannot return automatically, to the document of 
departure. 
Bi-directional Link 
The source anchor leads the user to a destination node, 
from where they can 'reverse the link', or in other words 
return to the node of departure. In Microcosm this occurs 
either, when both the source and the destination node stay 
active on the screen so that the user can jump from B back 
to A. or when. the destination document has replaced the 
source node (in Microcosm terminology: the viewer starts 
unlocked) but includes an anchor that connects back to it. 
The 'target" at the destination node is an attribute mostly 
used in situations where the destination is a text type of 
node. It represents the location within a text node, where 
the link points to: e.g. specific location, beginning of 
document. For example a link from the source anchor 'W. 
Thomson" to the specific location of a bibliographical 
reference in the bibliography type of text node. 
3.1.5 Types of Anchors 
The aspect of interest "Types of Anchors' is represented on 
the Pausanian map. in order to show what part of a node is 
linked to other information units. 
Anchored Text 
Any piece of text serving as a mark. emphasising (visible), 
or containing (invisible) a connection with another 
inRmnation unit; tier example the visible anchor 
'Democracy" which links to a text node about the 
Constitution in Ancient Athens. 
Anchored Graphic Object 
An3,, graphic element serving as a link to another 
information unit: for example a navigation sign on the map 
of the Acropolis. a sound icon at the end of a text node 
which links to a musical reconstruction from the Orchestra 
in the Athenian Agora. 
Anchored Graphic Selection 
A link made on a particular selection within a picture or a 
graphic e.g. a link made on an ancient wine cup (being part 
of a picture presenting six red-figure wine cups) connecting 
to another picture consisting of the bottom of that cup 
which depicts an ancient potter sitting at his wheel, the 
speed of which he controls with his knee and above him on 
a shelf there are some of his pots. 
All three types of anchors are attached to another set of 
attributes that represent the destination flag which is any 
possible indication at the source anchor for the type of the 
destination node. 
Destination Flag 
The destination flag will be shown on the Pausanian map 
when a source anchor is accompanied by any indication of 
its destination e.g. an icon: a video camera icon situated 
next to an anchor showing that the destination node is a 
video sequence; or any other way of indicating the 
destination such as a label appearing after a mouse action: a 
label stating "animation" which appears when the mouse is 
over an anchor; or a link preview: where the source anchor 
includes a preview of the destination node such as the node 
type and a name such as a title or a brief summary of the 
content. 
3.1.6 The Location of Anchors 
The aspect of interest 'location of anchors' includes 
attributes that show in the map, the exact position of each 
anchor in the source document: Paragraph No. Beginning 
of Document. Middle of Document. End of Document. It is 
a very important aspect of interest because it shows clearly, 
the frequency or concentration of links within a document. 
3.1.7 The Depth of Path 
The Depth of Path is an aspect of interest represented on 
the Pausanian Map in the form of steps. Those steps arc the 
attributes that identify the depth of the linking path. from 
the "surface" of the hypermedia document (tile source 
anchor) to its "bed" (the last linked destination node). For 
example: \[anchor AI\] in node A linked to Node B (1" 
Step) \[anchor B3\] in node B linked to node F (2"" Step) 
\[anchor F3\] in node F linkcd to node L (3'" Step), \[anchor 
A2\] etc. 
The Pausanian Map includes a visual representation of links 
traversed up to the Iburth step. If a linking path "pulls" the 
user further than the fourth step. then only the location of 
that last destination node is shown but not the 
representation of its attributes. 
3.2 EXOGENOUS ASPECTS OF INTEREST 
The second category of aspects of interest to the 
hypermedia author is the exogenous. The e.rogenous 
aspects of interest define the visual appearance, of a 
hypermedia document: 
Length of Document. Scrolling document (more than one 
screen page), Non-Scrolling document (one screen page). If 
the document is scrolling, it is represented in the Pausanian 
Map by the number of pages. Further attributes that are 
being considered include: Width of Text Block, Alignment 
of Text Block, and Nurnber of Fonts used in the 
Document. 
3.3. AUCTORIAL ASPECTS OF INTEREST 
The final category of the aspects of interest to the 
hypermedia author, is the auctorial I. The auctorial aspects 
of interest are concerned with the content's structure and 
I . We use the name Auctorial from the Latin word auctor which 
is the origin of the word author. What we mean by auctorial is. 
anything to do with the actual writer who writes his ideas in the 
form of hyperrnedia (presented in an electronic medium) and not 
in a linear way as the traditional author of the printed book. 
ccmaposition. We have classified these as: the status of 
association, the arrangement of cognitive layers, the content 
kermatism and the level of content fragmentation. 
3.3.1 The Status of Association 
A very irnportant aspect of interest for the hypermedia 
author is the association between the source anchor and the 
destination text node. The Pausanian notation includes 
eight different attributes which define the status of the 
association. These attributes are offered as predefined to the 
author, ( as well as to the content expert or the evaluation 
team) of the application. As well as the status of 
association we are considering including in the notation, 
Trigg's links taxonomy \[31\] which is a very explicit 
classification of attributes defining the semantic relation 
between nodes. 
Consecutive Association 
A consecutive association is the smooth content transition 
from the source node to the destination document. For 
example, the source node is about sculpture in ancient 
Greece and it discusses the different types of Korai (female 
standing figures) The third paragraph of the document 
explains briefly their characteristic elements and the 
differences between the Kore of the 6'" century BC and the 
Kore sculptured in tire 7" century BC. Within this 
paragraph there are two button links (visible) one on the 
anchor: "Kore of the 6'" century' and another on the anchor: 
"Kore of the 7'" century'. Both anchors lead the user to 
documents which discuss in detail the characteristics of 
each statue. If the user chooses to follow the link and read 
these documents they can easily compare their 
characteristics because tire source document has ah'eadv 
prepared them by irrtroducing the subject. 
Therefore there has been a smooth transition from the 
destination nodes to the source node and vice versa. The 
content of the destination document succeeds the source 
node. 
Renaissance Association 
Authors create a renaissance relation between two nodes, 
when they make a link to a document, that 'revives' the 
user's enthusiasm.fi~r learning more information about the 
corresponding subject. For example, the source document 
discusses the sculpture of the Parthenon where the author 
talks about its fiieze and metopai. In the text block 
discussing the fiieze of the Parthenon, the author has made 
a link on the word 'frieze' that connects to a detailed 
docurnent containing a description. The same strategy has 
been chosen for the word 'metopai'. Both anchors are 
visible (button) links. Since the source document is about 
'the sculpture of Parthenon' and includes links to 
information about the main architectural elements of the 
temple, it is building a renaissance relation where the 
author inclines the reader's mood to learn more about the 
anchored subjects. 
The distinction between the two examples given for the 
attributes of content relevance: renaissance relation and 
consecutive relation, is that in the renaissance relation, the 
destination nodes consist of information about different 
subjects that comprise the theme: Sculpture of Parthenon, 
(the user follows the link in order to learn more about the 
source subject), whereas in the consecutive relation the 
destination nodes consist of information on the same 
subjects (statues of Korai) but in different chronological 
periods. (the user-reader compares the information in the 
destination node, to the source node-a smooth transition 
from A to B) 
Umbilical-Cord Association 
An Umbilical-Cord relation is created when the destination 
node(s) complements the content of the source node. For 
example, the source document discusses the columns 
(kiones) of temples built in the Archaic period. The second 
paragraph of the text explains that the ancient temple's 
columns consist of parts such as the stylobate, the 
krepidoma, the capital and the entablature. All these words 
have been anchored by the hyperlnedia author as visible 
links, leading the user to destination nodes containing the 
description of each par! and its corresponding location on 
the column of an archaic temple. Since the destination 
documents complete the subject about the structu,'e of 
columns in the archaic temples, they create an umbilical- 
cord relation between the source node and its target 
docurnent(s). 
Subversive Association 
Hypermedia authors build a subversive relation when they 
connect two documents, and the destination document 
subverts the content of its source. For example the case 
where the source document is about the Parthenon's Frieze 
and its sculptured decor which represents the Panathenaic 
Procession. In the entire text the author concentrates on the 
interpretation of the sculptured representation and 
maintains the most popular theory which is the procession 
that accompanies the presentation of the robe of" the 
Goddess Athena. In the middle of this document, the author 
reports that there is another interpretation for the decor of 
the frieze, but it is not accepted by the experts. The words 
'another interpretation" are visible links connecting to the 
text reporting this alternative explanation which however 
subverts the globally accepted theory in the source 
document. In this case the author has created a subversive 
relation between the source and the destination node. 
Intrusive Association 
An intrusive relation is created when the destination 
document suddenly disrupts the coherence of the content. 
For example, the source document discusses the most 
famous ancient Greek sculptor Feidias and his master work 
on the gold statue of Zeus sculptured for the temple of" Zeus 
in Olympia. The author has chosen the words 'temple of 
Zeus in Olympia" (which are situated at the first paragraph 
of the source document) to be a visible anchor (button link) 
frorn which the user 'is given the opportunity' to follow the 
link and learn more information about this famous temple. 
The destination document happens to be extremely 
informative as two pages description of the temple 
including information about its architectural elements, the 
east and west metopies and what they represent (the labors 
of Heracles ). All tile main characteristic elements of the 
temple are visible anchors ( 15 button links) linking to 
other documents with more detailed information for each 
part of the temple i.e, its columns. The user started to learn 
about tile sculptor Feidias and his master work the gold 
statue of Zeus and ended up reading very detailed 
information about the temple of Zeus where the statue was 
situated, having previously passed from a text about the 
representation on the decor of the west fi'ieze. In this case 
the destination document has suddenly interrupted the 
reader from the content of tile source document and created 
an intrusive association. 
Magnetic Association 
The characteristics of a magnetic relation could be the same 
as the renaissance relation with the difference that the 
former brings possible danger. Let us take the same 
example as in the renaissance relation where the source 
document discusses the Pheidias" sculptures of the 
Parthenon, specifically the gold statue of the goddess 
Athena. Somewhere in the middle of the description the 
author refers to Pheidias" students who contributed in the 
construction of that most important temple of the ancient 
Hellenic civilisation. The words 'Pheidias' students' are 
visible anchors linking to documents of detailed 
information about each student. In this case the author has 
roused the reader's interest (magnetised) to learn about a 
subject completely different from the initial content. It is 
not a renaissance relation because there hasn't been any 
positive transition and the visible anchor leads the reader to 
a completely different context from the one in the source 
node. 
Association of Superiority 
A relation of superiority is created when the destination 
node is full of inlbrmation and overwhelms tile source 
document. For example the source document is about 
ceramic vase painting in the 7'" century BC, and discusses 
the Attic workshops and the technique for applying the 
decoration of the vases. In the third paragraph of the 
document the author reports a later technique (applied in 
the 6'" century BC). which was one of the two most 
common in the Attic workshops: the black-figure painting 
(the second most common technique was tile red-figure 
painting). The author explains that all vase-painting in the 
6'" century was based on the black-figure technique. The 
words 'black-figure" technique are a visible anchor, which 
leads the reader to a document containing a well composed 
detailed description on the actual technique, including 
information about different painters and their vase 
paintings. In this case the author by providing a visible link 
to such important and detailed intbrmation, overwhelms the 
content of the source document. 
Association of Discontinuity 
Content discontinuity is created when there is no relation 
between the destination document and its source. For 
example, suppose the source document is about 'the temple 
of Poseidon in Sounion'. In the second paragraph of that 
document there is a very brief reference to the columns of 
the Parthenon. The words 'columns of the Parthenon' are 
anchored as a visible link connecting to the document 
consisting of general information about the Parthenon. 
Since the source document is the temple of Poseidon and 
the destination node is about the Parthenon, there is no 
continuity in the content. If the destination document was 
about the columns of the Parthenon then there would be 
some continuity but not as much as in the umbilical-cord 
relation. 
Each of the above attributes are represented in the 
Pausanian map together with another attribute which states 
if the destination node has been composed by the same or 
different author. 
3.3.2 The Arrangement of Cognitive Layers 
Hypermedia authors, build cognitive jumps in order to offer 
the user what he anticipates from possible questions. They 
offer the reader, anticipated solutions to possible questions 
which arise, while reading the source document. We call 
these cognitive jumps the layers which are created by the 
reader's transition fi'om the informative source document 
to the destination node which focuses attention to a 
particular point of information. The author invents 
cognitive solutions that comprise tile necessary channels to 
enable the reader to attain their anticipated target. In some 
cases the author's intention to offer the anticipated, could 
be perceived by the reader as "a present". However, there 
are times when the readers get exactly what they expect. 
but there are also other times where they are given little or 
nothing that would meet their expectation. The hypermedia 
author is obliged to know his readers very well and plays 
the role of an augur who knows both, the person who is 
asking for an oracle as well as their problem. In other 
words, the reader and what they anticipate from the content 
in a source document, and the provided linked information. 
The cognitive channels or layers created by the author, 
resemble the oracles given by the augur (author) to the 
applicant (user). 
This mechanism of the creation of layers, if they have not 
been built carefully, will easily miss their target. 
The most important concern in that mechanism is what are 
the transitio~ls which create the layers. For example, if the 
reader studies a document about the Temple of Poseidon 
and follows a link from the visible anchor 'tile statue of 
Poseidon" to the actual picture of the statue, have they 
jumped to another layer? Probably not because the 
destination node is only a static (level of interaction: 
passive) graphic-picture which complements (explanatory 
relation) the anchor. Therefore the reader remains at the 
first layer. For example if a document is about the statue of 
the eponymous heroes in the Athenian Agora and there is a 
visible link on the anchor 'the Tribe of Leontis' (defined 
by the author as a magnetic association) leading to a 
detailed document about that tribe, then the author has 
guided the reader to another layer. We are still working on 
a clear classification of attributes that define these layers, 
so we have not yet included them as a representation on the 
map. 
3.3.3 The Level of Content Kermatism 
The kermatism of a hypermedia application is the 
segmentation and classification of the aggregate content 
into interrelated atomic subsets each of which comprise an 
individual subject. For example, the content of Athens in 
the Years of Perikles divided in chapters and sections such 
as The City, its structure and organisation, Social 
Organisation and day-to-day l(fe, Professions and Finance, 
The Athenian Constitution, Perikles, his life and mission, or 
a documetlt about the daily life in Ancient Athens and the 
information linked from anchors located within that source 
document. The hypermedia author by segmenting the 
aggregate content, gives each subset -section or subsection- 
an entity. They consider it as something discrete or 
individual within the entity of the aggregate content. 
However, this does not imply that those subsets do not 
intersect with each other. In a hypermedia application, they 
could exist within the content, as self-subsistent or sell'- 
created but they can also interrelate (be semantically 
linked). Some of the subsets, being either chapters sections 
or subsections, can supplement and complete a notional 
aggregation. By segmenting the content, authors can offer 
their best writing capability and meet their targets more 
easily. One of those targets may be the effective 
interrelation (linking) of subsets in order to achieve content 
coherence and comprehension. It is much more likely, that 
such targets will be met by the kermatism of the content 
(linked chapters, sections, subsections), rather than it, being 
a whole single linear unit. 
4.0 The Level of content Fragmentation 
We describe the fragmemation of a single h3permedia 
document, as all the typographic and conceptual 
interruptions which lead the user to a path of cognitive 
jumps. Both the kermatism and the fragmentation map will 
help the author to experiment with different strategies in 
order to see if readership improves or inclines. 
Unfortunately. due to space limitations we can not include 
their representations in this paper. (For details of. the 
Pausanian Kcrmatism Map, Pausanian Fragmentation Map 
and Pausanian Generic Links Map see HT'97 poster 
presentation: The Pausanian Notation \[ 14\] ). 
5.0 THE PAUSANIAN MAP, AN EXAMPLE 
The Pausanian Map shown in figure 2, represents the 
structure of a hypermedia document about "the statue of the 
eponymous heroes" in the Athenian Agora. 
From a first glance at the Pausanian Map of this particular 
document we can clearly see both the direction and 
distance o.fmovement. The author has included information 
guiding the reader along a linking path up to the fourth step. 
Research studies have shown that if readers are guided 
along more than four steps there is significant possibility 
that they will loose track of their current position \[16\]. If 
we \[bllow the linking path in reverse (from the fourth step 
to the first) we can see that the path started from the visible 
anchor "the Metroon'. The attributes of the destination node 
at the second step signify that the author has created an 
intrusive association between the destination node and its 
source. The hypermedia author has chosen, to describe the 
association between, the sou,'ce node and its destination, as 
intrusive. By definition, an intrusive association reveals a 
negative status. However, negative attributes between 
subsets, would not necessarily be negative for their 
hypermedia inclusion. This representation on the 
Pausanian Map, will help the author to visualise the 
thinking and structure of their ideas in order to examine 
how well they were composed within the hypermedia 
context. A suggestion in this particular example would be 
to avoid linking subsets that are not coherently associated. 
If the author particularly needs, to provide information 
about the Metroon, and make it available from that location 
(1" paragraph of the document, anchor: the Metroon), he 
could possibly change the selection of the visible Button 
type of link and make it, say an invisible Specific link. (A 
Local type of link would be a good choice only when the 
author wishes to make the information available from all 
the occurrences of the anchor "Metroon' in that document. 
However, if the link is described with a negative status of 
association, a local type of link is not recommended). 
The representation of the source document's linking 
structure shows the author that it contains six button links, 
two specific and one local (six visible and three invisible 
links). The length of the document is four screen-pages. If 
we consider its length, in combination with the number of 
links included, we can observe that there is no great level 
of fragmentation. 
The second paragraph of the document contains two button 
links one on the anchor 'Kleisthenic Reforms" and another 
on the anchor "Citizenship in Athens'. As represented on 
the map both visible anchors lead the reader to two 
different Graphic -Menu type of nodes. This means that 
while the users read the second paragraph of a seven 
paragraph document, they are "'offered" the opportunity to 
stop themselves from completing the reading of its content 
and to jump to a graphic-menu that contains links to 
completely different subjects. Again in this situation the 
author should either not link visible anchors to graphic 
menus in the beginning of the document or choose a more 
appropriate, invisible type of link. It is recommended 
however that the hypermedia author should not use these 
links at all, because both the content interruption, and the 
transition to completely difl~rent subjects, could mean the 
reader looses information taken from the source and 
become disorientated. 
Carrying on the examination of the structure in the fourth 
and fifth paragraphs respectively, two button links connect 
to pictures that complete the subject of the paragraph 
(explanatoly relation). The visible anchor 'The Eponymous 
heroes" connects to a picture which is accompanied by text. 
Undoubtedly, if carefully assigned, that strategy saves, a 
step and a layer, from the depth of links, and the 
arrangement of cognitive layers, respectively. The visible 
anchor 'preserved today' connects to another picture which 
is also accompanied by text. The map shows that this 
picture contains an anchored graphic object that links to a 
video sequence. However, this anchored graphic object is 
not accompanied by a destination flag, which means that 
the reader is not being informed by the author that they 
should expect khwtic information. It is recommended that 
all anchors connecting to either sound or kinetic 
information such as animation, video or movement in a VR 
environment, should be accompanied by a destination flag 
(i.e. a video camera icon or mouse label) to indicate what type 
of object is linked to the anchor \[19\]. Moreover. that video node is 
relatively long (6":35"'). Especially with teaching material the 
length of video and animation sequences should be restricted in 
proper length. Such long video nodes (more than 4 minutes) might 
prevent users \['or comprehending the content, reduce the 
concentration and will to study and give much more complex 
operation of software tools \[ 16\]. 
6.0 VARIABLES OF INTEREST- PAUSANIAN CHARTS 
Up to this point in the paper wc have introduced how the 
Pausanian notation could be used to represent a single 
hypcrmedia document in order to analyse its functionality 
or dynamics, and recommend possible improvements or 
alterations. 
As mentioned in the introduction, to visualise the structure 
of a group of hypcrmedia documents, we have defined a set 
of variables, (derived from the aspects of interest), which 
we call the rariables of interesr These are statistically 
represented in what we call "PausaHia~ Charts for 
representing the structure of a group of hypermedia 
documents" (Figure 3). Such variables include: Total 
Number of Visible Links. Total number of Generic Links, 
Maximum Depth of Path started from Visible Links. 
Maximum Depth of Path started from Generic Links. 
Concentration of Visible Links. Number of Associations of 
Discontinuity. Numbcr of Graphic-Picture Nodes. etc. (all 
variables are applied per hypermcdia document) 
Firstly. the author can select a group of documents lk)r 
investigation. (i.e. all the text nodes from the chapter Daily 
Life in the Pcriklcan Athens) , and then choose from our 
classification of the variables of interest, the ones they 
would like to represent on the Pausanian Charts, in order to 
examine their structure. Then a data table is generated and a 
series of provided types of charts, is recommended. The 
author can then choose one of the charts from which, the 
appropriate Pausanian Chart will be drawn, grouping 
together the selected variables and representing them 
statistically. By examining the Pausanian Chart, the author 
can visualise certain issues from the synthesis of all the 
selected documents, in order to perceive their structure at a 
glance. If they want to look more closely at a particular 
point in the chart they could apply the Pausanian Map on 
that particular node. This aspect of work is still under 
research. 
CONCLUSION AND FUTURE WORK 
By proposing the Pausanian notation we argue that it is 
time to start providing hypermedia authors with tools to 
help them find the problerns inherent in their designs 
themselves. By no means do we believe that a notation will 
solve all the problems. What we argue is that by defining 
the attributes that play a key role in the synthesis of a 
hypermedia structure and making them visually available 
to the composers of the application, possible problems and 
causes can be revealed, and the overall design and the 
performance of the application can be improved. 
In this paper we have introduced the Pausanian notation as 
well as its main features: the aspects of interest, the 
Pausanian Map, the variables of interest, and the Pausanian 
Chart. The major case study being used lbr evaluating and 
testing the notation, is Pausanias 2000, a h37)ermedia 
librao'for the ancient Hellenic WorM (The Attica Region). 
We are also currently evaluating our method on another six 
Microcosm and three Storyspace \[6\] applications. Due to 
space limitations, we cannot discuss in detail the results of 
our evaluation however, we have found that our approach 
has been justified and that. by using the Pausanian 
Notation, the author can quickly track and correct en'ors 
occurring throughout the structure. 
Our plans for future work are directed as follows: in the 
very near future we would like to have completed the 
classification of aspects (!f interest and rariables of interest 
as well as to improve the paper representation of the 
Pausanian Maps. In parallel, we will be working in the 
implementation of the Pausanian Notation for representing, 
examining and analysing the structure of Web based 
applications. Having achieved that. we will start 
automating the process of creating the Pausanian maps tk~r 
Microcosm applications, by adding the appropriate 
functions to the Microcosm Make-Link feature. The fact 
that Microcosm keeps all the link information stored in a 
linkbase file, makes it much easier to extract any 
information needed for producing the different Pausanian 
Maps. The implementation of the Pausanian method for 
representing the structure of a Web application, will be 
based on WebCosm (the Web version of Microcosm \[4\] ). 

REFERENCES 
1. Bieber. M. & Kacmar. C. Designing hypertext support 
for computational applications. Communications of 
the ACM. 38(8), 1995. 
2. Bolter. J. "Writing space, the computer, hypertext and 
the history of writing". Lawrence Erlbaum Associates, 
Inc., 199 I. 
3. Card, S.K. Robertson, C.G. Mackinlay, J.D. "The 
information visualiser, an information workspace". In 
proceedings of CHI "91 New Orleans, April 27-May 
2, 1991 181-188. 
4. Cam L, De Roure, D. Hall. W. and Hill, G. "The 
Distributed Link Service: A Tool for Publishers, 
Authors and Readers". In proceedings of The Web 
Revolution: Fourth International World Wide Web 
Conference, in The Web Journal I(I), O-Reilly and 
Associates, 1995. 
. Collier, G. H. Thoth II: hypertext with explicit 
semantics. Hypertext '87 Proceedings, pp. 269-290, 
New York, NY: ACM Press, 1987. 
6. Eastgate Systems, Inc. STORYSPACE: A Hypertext 
Tool for Writers and Readers, 
I) ttI~:llwww.eastgate.com/Storyspace.hlml. 
7. Dynamic Diagrams, Inc. Introducing MAPA, 
I) ttp:/Iwww,dynamicdiagrams.com/producls.htm. 
Davis, H. Hall, W. Heath. I. Hill, G. Wilkins, R. 
"'Towards an integrated information environment with 
open hypermedia systems". In ECHT'92: Proceedings 
of the Fourth ACM Conference on Hypertext, Milan, 
Italy (Nov.- Dec. "92), p. 181- 190, 1992. 
Furnas, G.W. "Generalised fisheye views". In 
proceedings of CHI 1986, 16-23. 
10. Garzotto. F. Schwabe, D and Paolini, P: 'HDM-A 
Model Based Approach to Hypermedia Application 
Design" ACM Transactions on Information Systems, 
I 1 (1) Jan. 1993, pp. 1-26. 
II. Ghedini Ralha, Cohn, A. "'Building maps of 
hypcrspace". WWW National Conference, Internet 
Muhimedia Information, July 6-8, Minho University, 
Portugal. 1995. 
12. Gloor. P. Cybertrees: "Tree-Shaped overview maps". 
In elements of hypermedia design, techniques for 
navigation and visualisation in cyberspace, Gloor P. 
cd. Birkhauser, 1997. 
13. Gloor. P. "Elements of hyperrnedia design : 
techniques for navigation and visualisation in 
cyberspace". Birkhaueser, Boston. 1997. 
14. Gouraros, N. "'The Pausanian notation, a method for 
representing the structure of a hyperdoeument". Poster 
Presentation ACM Hypertext "97, Southampton 1997. 
15. Grabinger, S. Dunlap, J. "'Links". In hypermedia 
learning environments: Instructional design and 
integration, Kommers, P. Grabinger, S. Dunlap, J. 
(eds). Lawrence Erlbaum Associates, Inc. 1996. 
16. Haga, H. Nishino, M. "Guidelines for designing 
hypermedia teaching materials". In Proceedings of the 
Conference in Educational Multimedia ED'95, 1995. 
17. Izakowitz, E, Stohr, E and Balasubramaniam, P: 
"RMM: A methodology for structured hypermedia 
design'. Comm. Of the ACM, October 1995, pp.34-44. 
18. Johnson. B. Shneiderman, B. "Tree-maps: A space- 
filling approach to the visualisation of hierarchical 
information structures". In Proceedings tbr 
Visualisation "91. San Diego, CA, 1991 284-291. 
19. Landow, G. "'The rhetoric of hypermedia: some rules 
lbr authors". In hypermedia and literary studies, 
Delany, Landow (eds). MIT Press, 1990. 
20. Landow, G. Hypertext 2.0: The Convergence of 
Contemporary Literary Theory and Technology 
Johns Hopkins University Press, 1997. 
21. Mackinlay, J.D. Robertson. G.G. Card. S.K. "'The 
perspective wall: Detail and context smoothly 
integrated". In proceedings for CHI '91, New Orleans, 
April 27 - May 2, 1991, 173-179. 
22. Maurer, H. Hyperwave: The Next Generation Web 
Solution. London: Addison-Wesley, 1996. 
23. Monmonier, M. "How to lie with maps". University 
of Chicago Press, Chicago 199 I. 
24. Mukherja, S. Folley, J.D. "Visualising the world wide 
web with the Navigational View Builder". In 
Proceedings for the Y" International World Wide Web 
Conference. Computer Networks and ISDN Systems 
27, no 6, 1075-1987. 
25. Mukherjea, S. Foley, J. Hudson. S. "Visualising 
complex hypermedia networks through multiple 
hierarchical views". In proceedings of CHI '95. 
Addison Wesley. 1995. 
26. Newman, W. Lamning M. (eds), "'Interactive 
systems design". Addison-Wesley. 1995. 
27. Robcrtson. G.G. Mackinlay, J.D. Card S.K. "Cone 
trees: Animated 3D visualisations of hierarchical 
information". In Proceedings lbr CHI "91, New 
Orleans, April 27 - May 2. 1991, 189 - 194. 
28. Rossi, G. Schwabe. D. Garrido, A.: Design reuse in 
hypermedia applications developmcnt. In proceedings 
of the ACM International Conference on Hypertcxt 
(Hypertext "97) Southarnpton, UK April 1997. 
29. Sarkar. M. Brown, M. "Graphical fish-eye views of 
graphs". In proceedings of CHI "92, Monterey, CA. 
May 3-7 1992, 83-9 I. 
30. Schwabe, G. Rossi and Barbosa, S: "Systematic 
Hypermedia Design with OOHDM'. Procceedings of 
the ACM International Conference on Hypertext 
(Hypertcxt "96) Washington. Marcll 1996. 
31. Trigg, R. A taxonomy of link types. In A network- 
based approach to text handling for the online 
scientific community. Pd.D. dissertation, Univeristy 
of Maryland Technical Report. TR-1346, November 
1983. 
32. Utting, K. Yankelovich, N. "Context and orientation 
in hypermedia networks". ACM Transactions on 
Office Information Systems 7, no. 1, January 1989, 
58-54. 
33. Zizi, M. "'Providing maps to support the early stage of 
design of hypermedia systems". In proceedings of 
IWHD'95 International Workshop on Hypermedia 
Design- 1995. 
