Most social networks and commerce websites are providing their users with friend and product recommendations. Ever wanted to add powerful user- and item-based recommendations to your social or commerce site? By integrating an advanced and proven recommendation engine (Apache Mahout) to do the heavy lifting, we can do this with Drupal too. This approach allows us to take advantage of the benefits of Mahout, including massive scalability and advanced computational algorithms, while keeping all of our data in Drupal. And we are not just limited to recommending users and products - this demonstration will include an example video recommender, and the possibilities are limitless!
In this session we'll be taking a closer look at Apache Mahout, the Recommender API for Drupal, and options for how to integrate the two, including a discussion of advanced customisation. We'll also discuss the implications and advantages of taking a loosely-coupled approach between the various systems, leveraging the Async Command module for processing the recommendations asynchronously.
- Recommender API
- Apache Mahout – investigating out-of-the-box algorithms and their application
- Integration with Drupal
- Asynchronous operation
- Making inferences
Feeding data back into drupal
- Views integration
- Tuning the results
Case study: SARACEN
- Integration with social networking data
- Profile word cloud
- Semantic web aspects
- Data aggregation
The research – carried out by Kendra Initiative – leading to these results has received funding from the European Union's Seventh Framework Programme (FP7/2007-2013) under grant agreement 248474 Socially Aware, collaboRative, scAlable Coding mEdia distributioN (SARACEN).
Daniel Harris spoke at DrupalCon Paris 2009 on Using Drupal for Media Asset / Content Management, Semantic Syndication / Promotion and Commerce and has a long history of speaking at conferences for Kendra Initiative.