How exactly would it be based on that? In other words, how would we measure similarity? All I can think of that's easy to measure is tags and perhaps upload date.
Besides that, I'm not sure viewing a resource, or even downloading it, really says a lot about whether you like it and want to see more of the same. (Unlike Youtube and Netflix and the like.)
Recommender systems are touch to make - they take either very precise fine-tuning or full-on machine learning, which would both be a lot of effort for very little benefit.