Material Detail

Improving Infoville XXI using Machine Learning Techniques

Improving Infoville XXI using Machine Learning Techniques

This video was recorded at Workshop on Machine Learning for User Modeling: Challenges, Edinburgh 2005. Infoville XXI is a citizen web portal of Valencia (Spain). This paper presents several approaches based on Machine Learning that help in improving the site. Three ways of improvement are taken into account: (i) user clickstream forecasting, (ii) user profiling by clustering and (iii) recommendation of services to users, being the last two techniques part of a methodological framework with general applicability that tries to be useful for a range of web sites as wide as possible. Results obtained with data sets from this web portal show that the most appropriate techniques for user clickstream forecasting become Support Vector Machines and Multilayer Perceptrons, whilst Adaptive Resonance Theory and Self-Organizing Maps appear to be the most suitable techniques for clustering. Final recommendation and adaptation of the recommender system is currently being developed by using Learning Vector Quantization and Reinforcement Learning.

Quality

  • User Rating
  • Comments
  • Learning Exercises
  • Bookmark Collections
  • Course ePortfolios
  • Accessibility Info

More about this material

Comments

Log in to participate in the discussions or sign up if you are not already a MERLOT member.