Planet Drupal

Recommender Module Performance Enhancement & Drupal for Data-intensive Computing

I've always wanted to build a cutting edge recommender system for Drupal as good as what Amazon offers. Google Summer of Code 2009 gave me the first chance to attack this task, and I developed the Recommender API module and helper modules that provides recommendation service based on users browsing history, fivestar ratings, product purchasing history, etc. After 2 years of application in the real world, I received many users feedback concerning performance/scalability issue of the modules, which cannot be fixed under the current PHP implementation -see why here-.

To solve the performance issue, I think the best option is to outsource the complex recommendation computation to Apache Mahout instead of using my own PHP implementation. I have submitted another GSoC application for 2011 to work on this. Hope it will get accepted so that I can get this done.

The second part of my GSoC 2011 application is to build a framework so that 3rd party programs, such as Apache Mahout, can easily exchange data with Drupal for data-intensive computing, such as computing recommendations. More details is discussed in my GSoC 2011 application. I hope this would facilitate more innovations on data-intensive computation with Drupal using 3rd party script/programs.

If you like these ideas, please support my application at

Drupal rocks, and let's make it rock more :D

"Related modules" block for -- Past and Future

Our research group at the University of Michigan has been working on the "related modules" block for for more than 2 years now. We have published 2 papers on this project so far:

1) Assessment of Conversation Co-mentions as a Resource for Software Module Recommendation. Will be presented at ACM Recommender System Conference'09

2) Conversation Pivots and Double Pivots. Presented at ACM Computer Human Interaction Conference'08

Announcing my GSoC 2009 project -- Making Drupal Smart: The Recommender Bundle

My Google Summer of Code 2009 proposal was accepted. The basic idea is to develop at least three modules based on Recommender API. For example, one module is to recommend Flash videos based on users' viewing history like in YouTube. A mockup screenshot is like this:

For more details and discussion, please go to

Pivots module recommendation system Google Analysis results

We developed 4 module recommendation algorithms and tested them on And we used Google Analytics and tracked the click-through rates. The overall click-through rate was 0.263%, co-occurrences 0.097%, relevance 0.141%, recency 0.114% and uniqueness 0.138%. The relevancy algorithm appeared to have the highest click-through rate, but it was only significantly higher than the co-occurrences algorithm.