int getTopSimUsers(gk_csr_t *adjMat, int user, gk_fkv_t *topUsers, int nsim) {
//to store the restart distributon of vertices
float *pr;
int i, j, count;
int iter = 0;
gk_fkv_t *pRanks;
pr = (float*) malloc(sizeof(float)*adjMat->nrows);
//initialize the restart distribution for user
pr[user] = 1.0;
iter = gk_rw_PageRank(adjMat, 0.5, 0.000001, 100, pr);
fprintf(stderr, "Iter: %d\n", iter);
count = 0;
//count the non-zero values in pr and sort them
for (i = 0; i < adjMat->nrows; i++) {
if (pr[i] > 0 && i != user) {
count++;
}
}
//printf("Found %d similar users.\n", count);
pRanks = gk_fkvmalloc(count, "store page ranks");
for (i = 0, j = 0; i < adjMat->nrows; i++) {
if (pr[i] > 0 && i != user) {
pRanks[j].key = pr[i];
pRanks[j].val = i;
j++;
}
}
nsim = gk_min(nsim, count);
gk_dfkvkselect(count, nsim, pRanks);
gk_fkvsortd(nsim, pRanks);
gk_fkvcopy(nsim, pRanks, topUsers);
free(pr);
free(pRanks);
return nsim;
}
void ComputeNeighbors(params_t *params)
{
int i, j, nhits;
gk_csr_t *mat;
int32_t *marker;
gk_fkv_t *hits, *cand;
FILE *fpout;
printf("Reading data for %s...\n", params->infstem);
mat = gk_csr_Read(params->infstem, GK_CSR_FMT_CSR, 1, 0);
printf("#docs: %d, #nnz: %d.\n", mat->nrows, mat->rowptr[mat->nrows]);
/* compact the column-space of the matrices */
gk_csr_CompactColumns(mat);
/* perform auxiliary normalizations/pre-computations based on similarity */
gk_csr_Normalize(mat, GK_CSR_ROW, 2);
/* create the inverted index */
gk_csr_CreateIndex(mat, GK_CSR_COL);
/* create the output file */
fpout = (params->outfile ? gk_fopen(params->outfile, "w", "ComputeNeighbors: fpout") : NULL);
/* allocate memory for the necessary working arrays */
hits = gk_fkvmalloc(mat->nrows, "ComputeNeighbors: hits");
marker = gk_i32smalloc(mat->nrows, -1, "ComputeNeighbors: marker");
cand = gk_fkvmalloc(mat->nrows, "ComputeNeighbors: cand");
/* find the best neighbors for each query document */
gk_startwctimer(params->timer_1);
for (i=0; i<mat->nrows; i++) {
if (params->verbosity > 0)
printf("Working on query %7d\n", i);
/* find the neighbors of the ith document */
nhits = gk_csr_GetSimilarRows(mat,
mat->rowptr[i+1]-mat->rowptr[i],
mat->rowind+mat->rowptr[i],
mat->rowval+mat->rowptr[i],
GK_CSR_COS, params->nnbrs, params->minsim, hits,
marker, cand);
/* write the results in the file */
if (fpout) {
for (j=0; j<nhits; j++)
fprintf(fpout, "%8d %8d %.3f\n", i, hits[j].val, hits[j].key);
}
}
gk_stopwctimer(params->timer_1);
/* cleanup and exit */
if (fpout) gk_fclose(fpout);
gk_free((void **)&hits, &marker, &cand, LTERM);
gk_csr_Free(&mat);
return;
}
