Trendy, design-oriented lifestyle products (also called life-like products) cause surprise,
create curiosity, are interesting and/or
amusing, or - in short - emotionally touch (potential) customers. Web presentation of such products is increasingly important, as
it became one important instrument to manage product and customer life cycles. Thus, attributes that apply to life-like products
should also apply to their websites, which implies that websites become life-like. The layout of life-like websites is self-determined
and adaptive to its environment, i.e. underlies an ongoing change in which web surfers (potential and regular customers) participate.
Self-determined layout does not only mean life-likeness of the concerning website; it also drastically simplifies the administration
of the website. Today’s life cycle of a website :: FIGURE 1 :: is characterized by too much
cumbersome manual interaction.
First and foremost, self-determined layout minimizes the scope of user-defined webpage layout and is twofold: creating an initial
webpage layout as well as adapting the layout based on user behavior. The web master can now focus on the core tasks, i.e. producing
content and defining the website structure. :: TOP ::
As traditional, layout-oriented web design tools do not meet the needs of life-like websites, one goal of the Polychaeta project is
to create a prototype of a new tool generation. This prototype provides functionality to define the website structure whereas a
layout generator is responsible for the initial webpage layout. Another goal is the implementation of a layout adaptation mechanism
which continually adapts the webpage layout by interpreting user behavior.
This project was launched as a student project in the summer term 2004, which resulted in a first prototype of a web design tool
and a layout adaptation mechanism based on the sugar scape model. The diploma thesis mainly focused on the implementation
of new adaptation concepts and on the redesign of the web design tool. :: TOP ::
· Web design tool
The web design tool :: FIGURE 2 ::
is used to define the structure and content of a website. The layout of the webpages is autogenerated
and continually adapted.
Traditional graphical user interfaces implement their functions with menus and buttons. But users find it difficult to learn the structure
of this kind of widgets. This becomes even worse, if the complexity of the provided functionality is high. To sum up, these approaches
require the user to explicitly choose a set of functions to complete a task.
One goal of our web design tool is to provide easy access to its features. We therefore decided to choose an approach which implicitly
does the right thing according to what the user expresses by his behavior. For these purposes, we have chosen drag-and drop
(in combination with a help assistant) as the primary means of user interaction. :: TOP ::
· Layout adaptation mechanism
Like the initial idea of the project itself, the layout adaptation mechanism is also biology-inspired.
Every webpage is regarded as a sugar scape which is divided into squares. Each square contains continuously growing sugar.
Furthermore, a square may be occupied by a webpage element (image or text element), also called agent. To survive, the agent
collects sugar from the squares. The collected sugar is eaten or may be stored into the agent's bag for later usage depending on
its sugar demand. As in real life, the sugar demand depends on the metabolism that is influenced by the size of the agent and its
sugar bag size. Below four sugar scape time units are shown, in conjunction with their equivalent websites.
:: FIGURE 3 ::
:: FIGURE 4 ::
:: FIGURE 5 ::
:: FIGURE 6 ::
Agents are sugar-driven. This obviously implies that the sugar distribution, i.e. which square provides how much sugar, guides the
agents. As a consequence, changing the sugar distribution will have an impact on the behavior of the agents.
We have outlined above that the webpage layout is adapted by interpreting user behavior. Internet users move the mouse and perform
mouse clicks when visiting webpages. This user behavior is regarded as a kind of "content preference", in other words, sugar
is distributed along the mouse paths and to the elements clicked on. This additional sugar has an obvious influence on the agent
behavior because agents need sugar to survive. Hence, the agents, i.e. the webpage elements, move and the webpage layout
changes. In summary, the way of visiting a webpage drives the layout adaptation.
:: TOP ::
During the summer vacations the application was migrated to the J2EE platform. The migration was primarily done for learning
purposes. As a side effect, the insufficient file-based persistence mechanism could be replaced by a database. Last but not least,
the resulting architecture provides a more integrated solution.
Several new adaptation concepts have been discussed and documented. Two of these concepts have been implemented: ”neighborhood
preferences” and ”rubbing stone”. ”Neighborhood preferences” finds elements on a webpage which have often been visited
(tracked by mouse-over) in sequence. A social behavior strategy is then applied to control the webpage elements. In our
case, we have developed a simple best-friend-strategy, making elements move towards the neighbor with the highest neighborhood
preference. The ”rubbing stone” concept transfers mouse-over events of a webpage to its neighboring (immediately preceding
or following) webpages.
To improve the comprehensibility and intuitiveness of the web design tool, the user interface was completely redesigned and a
help assistant was introduced (to mention but two measures). Usability tests combined with a survey showed that first-time users
had difficulties to understand the new way of user interaction. In a future extension of this project the usability concepts could be
the target of some rework.
:: TOP ::