There are many factors to consider for the successful
operation of web sites. One of the factors is the design of user
interfaces. User interface design is a very difficult and
complicated process, because it usually involves many aspects of
pages and page elements, and it needs to consider various users
and their goals using the systems. Until now, several methods
have been used generally for the design of user interfaces:
prototyping, user tests, and design principles.
Despite their widespread use, there are some limitations in
these existing approaches. First, prototyping and user tests
usually require a lot of time and manpower. Second, using the
design principles does not guarantee that the results will be
effective for all types of users. Third, there exist many different
types of users with different shopping behavior which cannot be
easily satisfied with a single design.
In order to address the above difficulties, this thesis
proposes the use of evolution strategy, an evolutionary
computing heuristic, for the design of user interfaces. For this,
an abstract model of a simplified Internet shopping mall is
constructed, which is optimized using the evolution strategy.
This thesis does not take into consideration the various
aesthetic aspects of user interface, because it is very difficulty
to quantify them in the heuristic method. Therefore, this paper
focuses only on the navigation scheme of the design.
The main contribution of this thesis can be summarized as
follows. First, this paper presented a method of user interface
design using evolution strategy, which can be used at lower
costs in less time compared with existing design approaches
such as user tests and prototyping. Second, this paper broadened
the application area of evolution strategy to user interface
design. Third, the experiments showed that the best user
interface design can differ for different types of users. This
implies that providing personalized user interfaces can possibly
improve user experience of Internet shopping malls.