Dataset::Dataset(std::string dataset_file)
:
ratings(read_mtx(std::ifstream(dataset_file), 1).csr()),
n_users(ratings.dim1()),
n_items(ratings.dim2()),
ica(generateICA(n_items)), // n_items * M
blooms(mult(ratings, ica, 1)), // n_users * M
sim(compute_sim()),
sim_dense(to_dense(sim)),
max_profile_size(0),
uniform_priors(n_items, 1),
empirical_priors(compute_item_priors()),
bloom_bits_priors(compute_bits_priors()),
ratings_dense(to_dense(ratings)),
rank(std::min(n_users, n_items)), // full rank
colMajorU(ratings.dim1() , rank),
diagS(rank),
colMajorV(ratings.dim2() , rank)
{
std::cerr << dataset_file << std::endl;
std::cerr << n_users << " users, and " << n_items << " items" << std::endl;
// if(graphs[dataset_no] != "null")//(2 == dataset || 5 == dataset || 7 == dataset)
// {
// std::cerr << graphs[dataset_no] << std::endl;
// CompressedFormat graph = read_mtx(std::ifstream(base + graphs[dataset_no])).csr();
// }
stack_assert(n_items == ica.dim1());
this->max_profile_size = 0;
for(index_t user = 0; user < n_users; user++)
this->max_profile_size = std::max(this->max_profile_size, nnz_row(user , ratings.starts()));
// SVD
smat mat = {this->ratings.dim2()/*!!*/, this->ratings.dim1()/*!!*/,
this->ratings.nnz(),
const_cast<index_t*>(this->ratings.starts()),
const_cast<index_t*>(this->ratings.index()),
const_cast<double*>(this->ratings.values())};
SVDVerbosity = 0;
SVDRec svd = svdLAS2A(&mat, this->rank);
std::cerr << "SVD computed!" << std::endl;
// i would take the pointer from SVDRec, but i'd have to remember calling free instead of delete[].
std::copy(svd->S, svd->S + this->rank, this->diagS.get_data());
// svd->Ut is row-major, so it is U col-major, save for V
std::copy(svd->Vt/*!!*/->value[0], svd->Vt/*!!*/->value[0] + this->ratings.dim1() * this->rank, this->colMajorU.get_data());
std::copy(svd->Ut/*!!*/->value[0], svd->Ut/*!!*/->value[0] + this->ratings.dim2() * this->rank, this->colMajorV.get_data());
svdFreeSVDRec(svd);
}
void read_matrix(idx *nv, idx *ne,idx **xadj, idx **adj, wt **ew, const char *filename){
std::string strfilename(filename);
if (endswith(strfilename, ".mtx")){
read_mtx (
filename,
nv, ne,
xadj, adj, ew,false,false,false);
}
else if (endswith(strfilename, ".bin")){
read_graph_bin(nv, ne,xadj, adj, ew, filename);
}
else {
throw std::runtime_error ("Reader is not available\n");
}
}
