I'm building a recommendation engine using ALS, as described in the title document.
I'm confused about a few points:
How should one interpret both X&Y, where X and Y are "factor vectors in the latent space" for both users and items, respectively. More specifically, once I factor Q (mxn) into X, which is mxr, how does one interpret the columns of X? Similarly, Y is an rxn dimensional matrix, what does each row correspond to? Here "r" is the rank of the reduced subspace. I'm looking for some sort heurestic interpretation for both X&Y.
How does one efficiently update this representation with new data? It seems that with each new piece of information, one must recompute both X&Y. Am I missing a trick?