StABLE: Making Player Modeling Possible for Sandbox Games
Date
2020-02-18
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
ORCID
0000-0002-9977-2662
Type
Thesis
Degree Level
Masters
Abstract
Digital games are increasingly delivered as services. Understanding how players interact with games on an ongoing basis is important for maintenance. Logs of player activity offer a potentially rich window into how and why players interact with games, but can be difficult to render into actionable insights because of their size and complexity. In particular, understanding the sequential behavior in-game logs can be difficult. In this thesis, we present the String Analysis of Behavior Log Elements (StABLE) method, which renders location and activity data from a game log file into a sequence of symbols which can be analyzed using techniques from text mining. We show that by intelligently designing sequences of features, it is possible to cluster players into groups corresponding to experience or motivation by analyzing a dataset containing Minecraft game logs. The findings demonstrate the validity of the proposed method, and illustrate its potential utility in mining readily available data to better understand player behavior.
Description
Keywords
Log analysis, Bag of Words, movement, motivation, experience, data mining, data analytics
Citation
Degree
Master of Science (M.Sc.)
Department
Computer Science
Program
Computer Science