12
Introduction
Paper D describes empirical scalability testing of the simulation platform pro-
posed in paper C; the method applied is factorial experimental design.
Paper E describes information gathering for enabling data mining in massively
multiplayer online games together with comparisons to web usage logs.
Paper F describes an algorithm for incremental proximal support vector classi-
fication using memoization to speed up classification with multiple classes
Paper G describes an algorithm for incremental and decremental proximal sup-
port vector classification
Paper H describes two parallel algorithms for incremental proximal support
vector classification using a heap-based tree topology. The first reads data
only on leaf-nodes (nodes = computational nodes) in the tree, and the
second reads training data on all nodes in the tree
Paper I describes empirical comparisons of the classifier proposed in paper F
with existing classifiers such as voted perceptron, c4.5, naive bayes, logistic
regression and SVM. Datasets used are data sets from the UCI Machine
Learning Repository (Blake and Merz [1998]), an actual web usage log and
simulated game usage logs using the simulator described in paper C.
Paper
User Rep.
User Pred.
A
X
B
X
X
C
X
D
X
E
X
X
F
X
G
X
H
X
I
X
Table 1.3: Relation - Papers and Thesis Topics