Pearson Similarity Question

I could not find any info or scientific paper about recommended data density for person similarity and matrix factorisation. I want to use them in my university project. Any hint with source would be helpful!

Hello! I think this paper might help.

In their testing they required a data density of at least 6%. Data density was the most important variable in their model of recommendation accuracy. With sparse data, matrix factorization performed better than collaborative filtering based on Pearson similarity.