A "micro browser" integrated into a handset allows displaying information to the user
and
accepting user requests. The basis for information representation is WML (Wireless Markup
Language), based on a datatype definition (DTD) for XML. The basic unit of WML is a card that
specifies a single user interaction screen. Navigation occurs between cards that are grouped into
decks. A deck is the top-most element of a WML document, presenting possible alternatives for
user interactions.
It is assumed that when the user logs on to a WAP gateway s/he supplies the address of a starting
page (WML deck) - a portal - and then performs interactions according to the cards in the deck. It
is desirable for the user to be able to find all the information s/he needs linked directly or with as
few links as possible from the starting deck. It is also preferable that favorite cards regularly
visited by the user can be reached through a shorter chain of links than cards that are only used
once in a while.
CONCEPTUAL SOLUTION FOR AGENT BASED MOBILE SERVICES
Our approach towards implementing mobile e-commerce services is based on using personal
software assistant agents. Software assistant agents were initially proposed as desktop agents for
support of user work with Personal Computers (PC) (Kozierok & Maes, 1993; Lieberman, 1995;
Maes, 1994; Van Dyke, 1999). However, we believe that personal assistants for mobile devices,
like cellular phones, significantly differ from software assistants for PCs. The main differences
are due to restrictions of wireless communications compared to wired communications (see
Introduction), limitations in user interaction with mobile devices, short connecting time of
customers and services in mobile communication, and a dynamically changing communication
context of users of mobile devices (for example, changing geographical location). All this puts
forward higher requirements of pro-activity, prediction of user needs, personalization and
delivery of precise information for the software assistant of a mobile device customer than for the
software assistant of a PC user.
In our solution each mobile device customer has his own software assistant agent which
represents his interests in the network. The assistant keeps a customer profile (preferences) and
analyzes and generates information to be made available to the customer by taking this profile
into account. This solution is presented in Figure 1.
Paper A
55