I am a current Ph.D. in the University of York supervised by Dr Daniel Kudenko and Dr Sam Devlin. My research interests are about "Game Analytics", more precisely, "Game Data Mining". The purpose of "Game Data Mining" is generally about building predictive models based on players' in-game behaviours in order to predict their future decisions. For example, it is common seen that players are leaving from a game for some reasons, but it is usually very hard to find out what game designs are correlated with their decisions. With "Game Data Mining", a predictive model could be built to link players' behaviours to their activities, therefore, their disengaging decisions can be predicted in advance. By the way, sometimes I also produce games for research.
Hanting Xie
Room B81, Foss Studios,
32 Lawrence Street
York, Yorkshire YO10 5FN
United Kingdom
Phone: +44(0)7472454233
Email: xiehanting@gmail.com
Ph.D. in Game Data Mining • Since 2013, 10
I started my Ph.D. in 2013 to do research in the area of game analytics supervised by Dr. Daniel Kudenko. Since then, I was mainly working on predicting players' disengagement and first purchases in two commercial games "I Am PlayR" and "Lyroke" which are both developed by "We R Interactive" (Recently purchased by Inspired Gaming). At the same time, I am developing a clone of Asteroids for my later research which is about data mining on game interface design.
MS.c. in Computer Science • 2011, 10 - 2012, 10
I was specialised in Machine Learning, Data Mining and High Performance Computing during my Master's degree. My final thesis is named "Efficient Matching of "If-Then" Rules".
Thesis Abstract:
The LCS (Learning Classifier Systems) are systems which have been broadly used
on different machine learning problems. This system is a rule ("If-Then Rules") based system.
Compared with other machine learning methods, LCS has a very slow run-time. And the most time
consumption part inside of it is "If-Then Rules Matching". In this study, the problem would be
optimized using different techniques, including: "HashMatch", "Recursion Optimization", "Object
Pool", "Parallel TTreeN" and "Input Tree + TTreeN". Among those, the "Input Tree + TTreeN"
method could even get a stable acceleration of more than 10 times faster than the naive matching
and it is developed originally from this thesis. The idea of it is to take the advantage of trees to
prune the duplicated inputs before performing the matching. Also, it transfers the traditional loop
comparison into the comparisons between the shapes of the trees. Benefits from those two features,
this approach offers a competitive improvement on its efficiency.
BS.c. in Computer Science • 2011, 10 - 2012, 10
I was specialised in Computer Technology and Application during my Bachelor's degree. My final thesis is named "DNA Sequence Alignment Using VHDL".
Thesis Abstract:
"DNA Alignment" is a useful technique in the area of medicine and biology.The
main idea of this research is to design and implement a hardware version of "Smith-Waterman
Alignment" on FPGAs using VHDL. A hardware version has its certain value over the software
design: firstly, hardware design has lower power consumption than a software design; Secondly, the
PCs are designed as general purpose machines, when a simple calculation is needed, several clocks
will also be cost for it, but hardware is designed to be dedicated into one problems, it’ll have higher
efficient than PCs.
Data Scientist• January 2017 – Present
Game Data Scientist Internship • June 2015 – Sep 2015
Part-time Data Editor • Nov 2014 – Sep 2015
Research Internship • Oct 2012 – Apr 2013
PhD Research • Jun 2015 – Present
Worked together with Dr. Pushmeet Kohli from Microsoft, Dr. Daniel Kudenko and Dr. Sam Devlin from University of York, this research is to compare several popular algorithms towards finding the best interface designs for individual players in a self-developed clone of ‘Asteroids’. I implemented the clone game with Unity and every algorithm for comparisons. The game is available on my personal website and the rest of research is still undergoing.PhD Research • Jan 2013 – Present
This research is conducted with Dr. Daniel Kudenko and Dr. Sam Devlin as the core part of my PhD’s research. Trends of player behaviours (e.g., retention and purchases) are important issues in game industry. However, most current methods are only available for specific games without generality. The main contribution of this research is to use frequency of game events as data representations to predict player behaviour trends. It provides better generality because no knowledge of any event but their frequency is needed. Two publications have been published.