def compute_scores(raw_data_dir=FLAGS.raw_data, data_dir=FLAGS.data_dir,
dataset=FLAGS.dataset, save_recommendation=FLAGS.saverec,
train_dir=FLAGS.train_dir, test=FLAGS.test):
from evaluate import Evaluation as Evaluate
evaluation = Evaluate(raw_data_dir, test=test)
R = recommend(evaluation.get_uids(), data_dir=data_dir)
evaluation.eval_on(R)
scores_self, scores_ex = evaluation.get_scores()
mylog("====evaluation scores (NDCG, RECALL, PRECISION, MAP) @ 2,5,10,20,30====")
mylog("METRIC_FORMAT (self): {}".format(scores_self))
mylog("METRIC_FORMAT (ex ): {}".format(scores_ex))
if save_recommendation:
name_inds = os.path.join(train_dir, "indices.npy")
np.save(name_inds, rec)
def recommend(raw_data=FLAGS.raw_data, test=FLAGS.test, loss=FLAGS.loss,
batch_size=FLAGS.batch_size, topN=FLAGS.top_N_items,
device_log=FLAGS.device_log):
with tf.Session(config=tf.ConfigProto(allow_soft_placement=True,
log_device_placement=device_log)) as sess:
mylog("reading data")
(_, items_dev, _, _, u_attributes, i_attributes, item_ind2logit_ind,
logit_ind2item_ind, _, user_index, item_index) = get_data(raw_data,
data_dir=FLAGS.data_dir)
from evaluate import Evaluation as Evaluate
evaluation = Evaluate(raw_data, test=test)
model = create_model(sess, u_attributes, i_attributes, item_ind2logit_ind,
logit_ind2item_ind, loss=loss, ind_item=None)
Uinds = evaluation.get_uinds()
N = len(Uinds)
mylog("N = %d" % N)
Uinds = [p for p in Uinds if p in items_dev]
mylog("new N = {}, (reduced from original {})".format(len(Uinds), N))
if len(Uinds) < N:
evaluation.set_uinds(Uinds)
N = len(Uinds)
rec = np.zeros((N, topN), dtype=int)
count = 0
time_start = time.time()
for idx_s in range(0, N, batch_size):
count += 1
if count % 100 == 0:
mylog("idx: %d, c: %d" % (idx_s, count))
idx_e = idx_s + batch_size
if idx_e <= N:
users = Uinds[idx_s: idx_e]
items_input = [items_dev[u] for u in users]
items_input = map(list, zip(*items_input))
recs = model.step(sess, users, items_input, forward_only=True,
recommend = True, recommend_new = FLAGS.recommend_new)
rec[idx_s:idx_e, :] = recs
else:
users = range(idx_s, N) + [0] * (idx_e - N)
users = [Uinds[t] for t in users]
items_input = [items_dev[u] for u in users]
items_input = map(list, zip(*items_input))
recs = model.step(sess, users, items_input, forward_only=True,
recommend = True, recommend_new = FLAGS.recommend_new)
idx_e = N
rec[idx_s:idx_e, :] = recs[:(idx_e-idx_s),:]
# return rec: i: uinds[i] --> logid
time_end = time.time()
mylog("Time used %.1f" % (time_end - time_start))
ind2id = {}
for iid in item_index:
uind = item_index[iid]
assert(uind not in ind2id)
ind2id[uind] = iid
uids = evaluation.get_uids()
R = {}
for i in xrange(N):
uid = uids[i]
R[uid] = [ind2id[logit_ind2item_ind[v]] for v in list(rec[i, :])]
evaluation.eval_on(R)
scores_self, scores_ex = evaluation.get_scores()
mylog("====evaluation scores (NDCG, RECALL, PRECISION, MAP) @ 2,5,10,20,30====")
mylog("METRIC_FORMAT (self): {}".format(scores_self))
mylog("METRIC_FORMAT (ex ): {}".format(scores_ex))
return