Brief Overview of my Research 2002Oct25

1
Brief Overview of my Research2002-Oct-25

• Amund Tveit, PhD student
• Department of Computer and Information Science
• Norwegian University of Science and Technology
• amund.tveit_at_idi.ntnu.no
• http//www.idi.ntnu.no/amundt/
• 47 4 162-6572

2
Background

• Growing interest in Data Mining, based on
  experience from
• Process Databases and Visualization for prod.
  managers (1990s)
• Key Production Parameters (Elkem Thamshavn)
• Statistical Models (Elkem Research)
• Large-scale Call Center Solutions (1996/97)
• Parallel Systems
• Telecom Industry (IBM Canada, Bell Canada,
  Stentor)
• Customer Analysis (of up to 15 million customers)
• Acceleration of Web Browsing (2000/02)
• Prediction and Prefetching (Agentus)

3
Research Problems

• How to Perform Efficient and Automatic
  Personalization of Large-Scale Multiparticipant
  and Interactive Online Systems?
• How to predict near future actions of users in
  such systems?
• Data Mining and Statistical Methods
• Scalability
• How to make use of those predictions to improve
  and adapt the system in order to make the user
  (e.g. paying customer) more satisfied with the
  system?
• Recommender Systems (Collaborative Filtering)

4
Example Systems

• I am examining 2 types of example systems
• Traditional and Mobile Internet (WWW)
• News Service Example
• Allready large and increasingly growing industry
• Massively Multiplayer Games
• Current Market Size US 300-400 Million
• Anticipated growth to US 2.7 Billion in 2007
• Norway among the pioneers! (Funcom, RazorWax)

5
Methodologies

• Prototyping/Proof of Concept (from 2000)
• Agent-Oriented Software Engineering (from 2000)
• Algorithms and Theory (from 2001)
• Parallelization
• Machine Learning (e.g. Support Vector Machines)
• Data Mining (e.g. Preprocessing and
  Classification)
• Stochastic Simulation (from 2002)
• Mobile Users browsing a News Site
• Massively Multiplayer Role Playing Game
• Empirical Methods (from 2002)
• Validation

6
Conclusion

• Hope to achieve 3 main goals within my PhD
  scholarship period
• Have developed Methodologies and Technology for
  efficiently Accelerating User Browsing
  Performance
• Likelihood of achieving high
• Enable useful Data Mining for Massively
  Multiplayer Games
• Likelihood of achieving medium (high..
  Optimistically? ?)
• Compare and show relations between the above 2
  goals
• Likelihood of achieving high