Roadmap of the next update (2010-09-13): click here
This module, once enabled, provides 4 blocks:
1. Users who voted similarly as you
2. Recommended nodes from users who voted similarly as you
3. Other nodes voted similar to this one
4. You might also interest in these nodes based on your previous votes
The first 2 blocks are computed using the User-to-user algorithm. The first block returns a list of users who have made similar votes as the current user. The second block then recommends a list of nodes to the current user that were highly voted by those users from the first block. For detailsof this algorithm, please refer to: http://portal.acm.org/citation.cfm?doid=192844.192905
The other 2 blocks are computed using the Item-to-item algorithm. Block (3) returns a list of nodes in the sense of "users who liked this node also liked". It's different from block (2) because block (2) is personalized towards the current user, where as block (3) is not personalized. Block (4) returns a list of nodes similar to the nodes you voted highly. The "item-to-item" algorithm is also used by Amazon.com. Please refer to http://portal.acm.org/citation.cfm?id=642471 for more details about the algorithm.
This module requires http://drupal.org/project/recommender. After installation/configuration, please go to admin/settings/recommender and generate the recommendations, then enable the 4 blocks as needed.
Note:This module only supports Fivestar, not any other VotingAPI-based modules. My design goal is to simplify configuration as much as possible by targeting on Fivestar only. To add recommender feature to Vote Up/Down or Plus 1, please install Points Voting Recommender. To support other VotingAPI-based modules, you can create new customized modules based on this one.
This module is different from CRE in that: 1) CRE only supports SlopeOne algorithm, not User-to-user or Item-to-item, 2) CRE supports generic VotingAPI-based modules, but requires more configuration.
The developer wasn't able to test this module on a large scale production site, so please feel free to submit issues if it doesn't work as expected on large scale sites.
This module is sponsored by Google Summer of Code '09.