Matrix factorization on implicit feedback score. How should i do it?

Question

I am trying to build a group recommender system.

i have feedback score in this format

1.023400000000000087e+00 1.005066999999999950e+02 5.539999999999999786e-02

where the last two digits after the plus sign depict user and item individually. HOW should i pass this information through the recommendation engine to get meaningful results ?