8
Introduction
RQ3: How can classification algorithms be used for customization of cyberspace
services, e.g. mobile commerce and massively multiplayer online games?
1. Why is classification appropriate for customization of cyberspace ser-
vices?
2. How can classifier algorithms scale to handle a large amount of cus-
tomers and customer personalization types?
3. How can classifier algorithms handle changes in data over time?
4. How to evaluate classification performance of the proposed classifier(s)
compared to state-of-the-art classifiers, and what are the results?
Research Question
User Rep.
User Pred.
MRQ
X
X
RQ1
X
RQ2
X
X
RQ3
X
Table 1.1: Relation - Research Questions and Thesis Topics
Research Field
Since the Electronic Commercial Agents project goal is about applying agent
technology in various e-commerce settings, the research field of study can be con-
sidered a part of the Web Intelligence (WI) field (Zhong et al. [2003, 2000]).
WI is interpreted as the application of Artificial Intelligence (AI) in an Internet
context. Software agents belongs to the Distributed AI field and classifier algo-
rithms belongs to the Machine Learning field, both considered subfields of AI
(Moulin and Draa [1996]; Mitchell [1997]). Web Intelligence and its subtopics
considered relevant for this thesis will be further described in chapter 3.
1.4
Contributions
The work in this thesis has been a combination of the design of software architec-
ture and algorithms, performing experiments, programming and developing and
running computer simulations. This work has been documented in the papers in
Part III.