SoC Application - Rcomendation Engine

Well, I submitted my application on time. But I was in the middle of finals and preparing for graduation so my application was not as fully thought out as I would have liked. Since no one has commented on my app in awhile I figured I would but forth my newer thoughts.

I orginally purposed a collaborative filtering approach that 'learned' what the user like based on what they read.

That is really hard to do. Furthermore, I was having a hard time coming up with a good solution to do that.

So, instead i would like to purpose a collaborative filtering approach called Slope One Predictors (http://www.daniel-lemire.com/fr/documents/publications/lemiremaclachlan_...)

It is updateable on the fly, effiecent and little information is required for the user. In other approaches, collaborative filtering alogrithms have the user fill out a survery to extract details about that user. Using this approach, only the simple votes are necessary. This paper outlines 3 related but different approaches which will allow Drupal admins to pick (presumally based on their users feedback). This is a simple approach which can later be agumented by a Information Filtering agents.

This algorithm is much easier to implement and the authors show that their alogrithm produces MAE scores similar to PEARSON algrithm even while being much simplier and much more efficient to query.

I am very interested in this project. I have already written my first module and I have a pretty decent grasp of Drupal API though it seems pretty extensive.

I hope that the mentors will continue to consider me for this project. I believe I will pursue this with or without SoC but my time spent on it will be far less. This project would be my only job over the summer.

~Scott