An Explicit Model For Tailor-Made Ecommerce Web Presentations

An explicit model for tailor-made eCommerce
web presentations
S. G. Loeber1
, L. M. Aroyo1
, L. Hardman2
1
TU/e, Computer Science, P.O. Box 513, 5600 MB Eindhoven, The Netherlands,
telephone:+31.40.247.5154, fax: +31.40.246.3992
2
CWI, P.O. Box 94079, 1090 GB Amsterdam, The Netherlands,
telephone:+31.20.592.4147, fax:+31.20.592.4312
Abstract. This paper focuses on how the setting-related factors influ-
ence the content selection, the layout creation and the presentation gen-
eration for (eCommerce) web-based presentations. It pays special atten-
tion to those factors which help the user play rhetorical roles. We have
integrated our theory in the Standard Reference Model for Intelligent
Multimedia Presentation Systems (SRM-IMMPSs). By doing so, both
the user and the designer will benefit. The user will be able to interact
and communicate in a more natural way and the designer will be able
to predict, limit the number of presentations plus perform more precise
usability evaluations.
1 Introduction
In the current competitive business environment, providing meaning and satis-
faction to customers is of great importance in order to remain in business. A
web site can be of help in many ways: it can be used to attract the customer’s
attention to (new) information, products and services; it can be used to persuade
customers to purchase a certain product directly or to live according to a certain
lifestyle; and it can be used to offer reassurance and service to keep the customer
satisfied and loyal [5].
In order to automatically generate web-based presentations which help users
to fulfill their needs and to become loyal customers, a complex model is needed.
This model should not only contain the information about the user, but also
about the setting in which the presentation will be presented. Our model, de-
rived from the Standard Reference Model for Intelligent Multimedia Presentation
Systems (SRM-IMMPSs) [1], introduces an Optimization Layer which guides the
creation of presentations in such a way that it helps users understand how to
react to or to interact with the information provided. In other words, this type
of presentations help users recognize the rhetorical role [3] they are asked to
play and helps them to understand the purpose and the content of the presen-
tation. When a user plays a rhetorical role, for example a ”teacher” looking for
study books on Amazon.com, she presents a version of herself that she deem
appropriate for the time (3 months before her course starts), place (web site

that sells books) and situation (finding cheap books that students will like) [8].
The combination of the user, time, place and situation is called a setting.
To sum up: information about the setting and rhetoric (which role to play
in order to fulfill a need through interaction with the web-based presentation)
within our model is necessary because every course of action from the users is
not only driven by their needs and goals, but is also dependent on its virtual
and social circumstances (which content, layout, interaction level is provided
plus which device, platform, network speed is being used).
In this paper we focus on the setting-related factors that influence the content
selection, the layout creation and the presentation generation for a web site. We
aim at a model that facilitates the process of rhetorical role-playing presentation
generation. In section two, we introduce the characteristics of an eCommerce
oriented setting and the rhetorical roles users play within the different types
of web-based presentations. In section three, we describe the SRM-IMMPSs. In
section four, we propose an Optimization Layer and corresponding adjustments
within the IMMPSs architecture. Finally, we present discussion and conclusions
of the proposed approach.
2 The characteristics of an eCommerce oriented setting
In order to describe an eCommerce oriented setting, we focus on two main char-
acteristics: the web site and the visitors. In the following two subsection we will
briefly describe the influence of the web site and the visitors in the process of
automatically generating an adaptive tailor-made web presentations.
2.1 The web site
The main purpose of a web site is to inform, to persuade or to entertain its target
audience. The purpose of an eCommerce oriented web site is to persuade its
target audience in such a way that they will buy, use or adjust to the information,
products, service/lifestyle offered.
Each web site type is either based on utilitarian values or hedonic values.
Web sites with utilitarian values are designed to help the visitor perform certain
tasks easily and quickly. Their design is all about functionality. They are realized
as a feature-based implementation such as search engines, structured lists and
overviews/site maps. Web sites with hedonic values are designed to entertain the
visitor. Their design is all about pleasure, excitement and they offer a uniquely
webby experience with characteristics such as association, image, experience and
pleasure.
We consider eCommerce oriented web sites to be mainly of an electronic shop
(eshop) type, where the purpose of the web site is to persuade people to order
and/or buy information, products, lifestyles, dreams, etc. They can have either
utilitarian values, where the visitor uses, for example, a search engine to find the
products and click on a button to order that product or hedonic values, where
the visitor for example plays a game or reads a story to get a reduction or a free
sample.

2.2 The people who visit the web site
When people communicate with each other they communicate through a series
of [rhetorical] roles that they assume appropriate to their rhetorical purposes
[2]. Their rhetorical purpose is to find information written in such a way that it
helps them to fulfill their needs. Therefore they will prefer information written
in a informative, persuasive or entertaining way. We presume that users also
play different roles when they react to or interact with a web site. In order
to incorporate rhetorical role playing into the electronic medium, we introduce
Persona Theory [3]. Persona Theory provides guidelines how to create roles
that define the content, the tone and the ways to communicate with the web
site and its audience. This is achieved by introducing Personas. Personas are
behavioristic style templates. Each Persona shows how to achieve its goal and
perform its tasks within a certain setting, in this case, a web site with a certain
purpose and a certain way of interacting with the visitor.
There are two types of Personas. On the one hand the Author Persona, who
is the voice (of the designer/company) that speaks through the web site. On
the other hand, the User Persona, who is a voice that explains the role the
actual user is asked to play when entering the web site. User Personas can be
grouped into do-ers, readers and viewers [10], based on their given needs and
goals. (We use the term do-ers, rather than Zeldman’s users in order to avoid
confusion with the general interpretation of the word user). Do-ers are people
who use tools to accomplish a task. They want to find a product (so they can
buy it) or find/compare/verify information. They want to carry out their task
as fast as possible and with as little effort as possible. Viewers are people who
seek entertainment. They want to be surprised, seduced or guided because the
journey is often more important than the end result. They will take their time
and enjoy the web site’s ’flashy’ appearance. Readers are people who turn to
a web site as they might turn to a book or an article. They want to read in
order to find in-depth information or to find relaxation/excitement. Do-ers are
more attracted to eshops with utilitarian values. Viewers like to be entertained
or want to experience a certain life style on eshop with hedonic values. Readers
will prefer an eshop which looks like a story, where for example the visitor can
buy products following the example of a celebrity or one of the characters within
this story.
Following this reasoning, we can conclude that there should be a link between
the purpose of the web site and the type of audience when developing Personas
in order to help users to fulfill their needs and become loyal customers. In the
following sections, we first shortly describe SRM-IMMPSs, then we present the
Optimization Layer which makes it possible to generate setting and rhetorical
role-playing dependent presentations.
3 The SRM for IMMPSs
A common approach for presentation generation is realized with the help of
retrieval, creation and generation systems. An example of this is the Standard

Reference Model for Intelligent Multimedia Presentation System (SRM-IMMPS)
[1], which deals with the dynamic creation and generation of multimedia pre-
sentations. It is a broadly applicable model that describes the generating of
presentations including web-based presentations. Therefore our model is based
on SRM-IMMPS. In the next section web will introduce an Optimization Layer,
which lies between the Generation Process and the Knowledge Server (see fig
1). First we will show the IMMPS architecture and shortly describe the different
expert modules within the Knowledge Server.
Fig. 1. The Standard Reference Architecture for IMMPSs
Figure 1 shows the IMMPS reference architecture. Its conceptual design re-
flects a modularization of the generation process into five layers: Control Layer,
Content Layer, Design Layer, Realization Layer and Presentation Display Layer.
To gain knowledge, the layers can exploit explicitly encoded knowledge provided
by a separate Knowledge Server. The Knowledge Server is composed of four
knowledge sources: Application Expert, Context Expert, User Expert and Design
Expert. The experts are represented as shared resources rather than as local or
private resources.
The User Expert maintains the knowledge of the IMMPS’s user model. It
may include representations of the goals and plans; physical or mental abilities;
attitudes and preferences; knowledge and beliefs of the user or group of users.
The Application Expert provides the IMMPS with application specific knowl-

edge. The data can be retrieved from different sources, such as web sites, databases
etc. Within this component the data for the presentation of the web site is stored.
It is also assumed to carry out tasks, such as interfacing with the application
system(s), data format conversion, provision of data and data characterization.
The Context Expert serves as a container for knowledge concerning the con-
text. The context refers to the device used, the platform, the network speed and
all factors that can influence the final presentation without being part of the user
characteristics. Its task is to construct a generation context (a representation of
what has been generated so far and what adaptivity rules have been used) and
a presentation context (a representation of what has been presented to the user
so far and how she interacts with this presentation).
The Design Expert contains the design rules on how to build a presentation.
It complements the other experts in the sense that it provides a container for all
other knowledge which is relevant for decision making in an IMMPS. The design
constraints and rules are organized per device model and media/modality model.
4 Introducing an Optimization Layer for eCommerce
In order to be able to automatically generate adaptive, tailor-made presenta-
tions, a setting-driven and rhetorical role-playing approach is needed. Therefore
an User Model Expert with Target Audience and Visitor Profile knowledge, a
Context Model Expert with Context and Usage Profile knowledge, a Discourse
Expert, a Design Expert and an Application Expert within the Knowledge Server
are necessary to provide all the information about the setting. In order to come
up with the best presentation for a specific user within a certain setting, the
data from the experts needs to be retrieved by an Optimization Layer. This
layer filters the data in such a way that the Motivation, Ability and Opportunity
factors have the highest possible values. Afterwards, it sends the processed data
to the Generation Process so that a rhetorical role-playing and setting oriented
presentation can be generated (see fig. 2).
4.1 Updates on the Knowledge Server
The User Expert is changed into User Model Expert and contains a Target Au-
dience Expert and a Visitor Profile. The Target Audience Expert represent the
target audience’s motivation, ability and opportunity [6], [4]. The Visitor Profile
contains knowledge about a particular visitor and additional information to the
Target Audience Expert, for example user preferences. Our User Model Expert
contains not only information about the presumed user model, but also infor-
mation directly retrieved from the actual visitor. By generating presentations
specifically meant for a certain audience, we hope to get a more precise User
Model Expert. The SRM-IMMPS has either one of these information sources
and targets as much visitors as possible.
The Context Expert is changed into Context Model Expert and contains a
Context Expert for general information and an Usage Profile for specific infor-
mation about the way and product the user uses to view the presentation. The

Fig. 2. An Optimization Layer within the SRM for IMMPSs
added value of the Context Model Expert compared to SRM-IMMPS’s Context
Expert is that it gives a more precise model on initially processed, integrated
information.
A Discourse Expert [9] is added. This Expert guides the generation of the
presentation structure using discourse models such as Rhetoric Structure Theory
[7], a pedagogical model or a marketing oriented model. It is based on the purpose
of the web site and gives a recognizable ’flavor’ to the presentation, which shows
how users should play their roles. This ’flavor’ can be the color of the brand or
a certain way of writing that distinguishes one web site from another within the
same application area.
The Application Expert and the Design Expert remain the same. They are
both used in the process of retrieving and creating the content for the presenta-
tion.
4.2 The Optimization Layer
In figure 2 we present the Optimization Layer which consists of three modules:
Motivation, Ability and Opportunity. These modules are necessary for processing
the data from the Knowledge Server in such a way that a presentation can be
generated based on a rhetorical role-playing and setting-driven approach.
The Motivation module is responsible for the filtering of information about
the target audience characteristics, such as the need, goal, cultural values in

combination with the purpose of the web site. It guides the generation of a pre-
sentation relevant to the target audience. It motivates the visitor to explore the
web site and fulfill her need. It contains information of personal relevance. It
keeps track of the values consistency (culturally held beliefs), needs and goals.
It also assures a low customer risk and controls a moderate inconsistency with
attitudes of the user. In order to gain this knowledge, the Motivation module
collaborates with the Target Audience, Application and Discourse Expert mod-
ules.
The Ability module processes the information about the user characteristics,
such as cognitive style, (dis)abilities, level of experience and knowledge. Its goal
is to ensure that the generation process results in a presentation suited for a
user with (dis)abilities, such as level of knowledge, level of experience, type
of cognitive style, motor-handicapped/poor-sighted, etc. In order to gain this
knowledge, the Ability module collaborates with the Target Audience and the
Design Expert modules.
The Opportunity module processes the characteristics of the environment
and characteristics on the presentation of the web site, such structure, structure
of content. Its goal is ensure that the generation process results in a presentation
that is adapted to the environment. The layout of the web site presentation is
improved and can be viewed by using that certain device, platform, network
speed, etc. In order to gain this knowledge, the Opportunity module collaborates
with the Context, the Application and the Design Expert modules.
5 Discussion and conclusions
This paper introduces issues that originate from Persona Theory and Situated
Action framework. Rhetorical role-playing and a setting-driven approach can be
helpful on the side of the enterprize and the customer within an eCommerce
setting. Based on this we offer an explicit model for a tailor-made web site
presentation in eCommerce web site setting. Following the advances on SRM-
IMMPS research with respect to dynamic creation and generation of content
presentations, we build upon the SRM by adding an additional Optimization
Layer. This way, we aim at generating meaningful and satisfactory presentations,
for visitors who recognize the role they need to play in order to find what they
are looking for.
The power of the Optimization Layer is twofold. On the one hand, the Opti-
mization Layer processes the data from the Knowledge Server experts and sends
it to the Generation Process in order to guide the generation of the most ap-
propriate presentation (setting, rhetorical role-playing adaptive) for the current
visitor. This makes it easier to perform precise usability evaluations. On the
other hand, the number of different presentations from the same web site is not
infinite. This will not only decrease the time for automatic generation, but it
will also make it easier to trace the number of scenarios within a certain setting.
This way a further user adaptation based on the Visitor Profile is facilitated. A
possible disadvantage of the approach is the fact that the web site presentation

is no longer suitable for visitors who are not part of the target audience, but who
are potential customers. It could also be the case that the information exchanged
among the Knowledge Experts conflicts.
To summarize: altering the SRM from a Situated Action and Persona Theory
perspective, provides means for satisfaction-level evaluation of the presentation
within the process of web design. This is an additional step to fulfilling various
customer needs, within, but not restricted to, eCommerce settings.
References
[1] Bordegoni, M., Faconti, G., Feiner, S., Maybury, M., Rist, T., Ruggieri,
S., Trahanias, P., and Wilson, M. A standard reference model for intelligent
multimedia presentation systems. Computer Standards and Interfaces 18, 6-7
(December 1997), 477–496.
[2] Coney, M. B. Technical readers and their rhetorical roles. IEEE Transactions
on Professional Communication 35, 2 (June 1992), 58–63.
[3] Coney, M. B., and Steehouder, M. F. Role playing on the web: Guidelines
for designing and evaluating personas online. Technical Communication 37, 3
(August 2000), 327–340.
[4] Hoyer, W., and MacInnis, D. Consumer behaviour, 2 ed. Houghton Mifflin
Company, 2001.
[5] Kobsa, A. J., and Pohl, W. Personalized hypermedia presentation techniques
for improving online customer relationships. The Knowledge Engineering Review
16, 2 (2001), 111–155.
[6] Loeber, S. G., and Cristea, A. A www information-seeking process model.
The 8th International Conference of The International Society for the Study of
European Ideas, July 2002. To appear in workshop CBMO 2002.
[7] Mann, W., Matthiessen, C., and Thompson, S. Rhetorical structure theory
and text analysis. ISI Research Report ISIIRR-89-242, University of Southern
California, Information Science Institute California, November 1989.
[8] Suchman, L. A. Plans and situated actions: the problem of human-machine
communication. Cambridge University Press, 1987.
[9] van Ossenbruggen, J., Hardman, L., and Rutledge, L. Hypermedia and
the Semantic web: A research agenda. Tech. Rep. INS-R0105, CWI, 2001.
[10] Zeldman, J. Designing your audience. A list apart: for people who make websites
by Jeffrey Zeldman and Brian M. Platz, 1999-2001.

An Explicit Model For Tailor-Made Ecommerce Web Presentations

More Related Content

Similar to An Explicit Model For Tailor-Made Ecommerce Web Presentations (20)

More from Martha Brown (20)

Recently uploaded (20)

An Explicit Model For Tailor-Made Ecommerce Web Presentations