def add_new_article(self, line):
article_id = int(line.split(" ")[0])
if article_id in self.bad_articles:
return -1
if article_id not in self.articles:
try:
article = to_vector(line)
except IndexError:
print("Skipping line, weird formatting.." + str(article_id))
self.bad_articles.add(article_id)
return -1
self.k_articles[article_id] = dict()
for a in self.articles.keys():
value = self.exponential_kernel(article, self.articles[a])
self.k_articles[a][article_id] = value
self.k_articles[article_id][a] = value
self.articles[article_id] = article
self.k_articles[article_id][article_id] = 1
return article_id
def add_new_article(self, line):
article_id = int(line.split(" ")[0])
if article_id in self.bad_articles:
return -1
if article_id not in self.articles:
try:
article = to_vector(line)
except IndexError:
print("Skipping line, weird formatting.." + str(article_id))
self.bad_articles.add(article_id)
return -1
self.articles[article_id] = article.reshape([1, 6])
return article_id
def add_new_article(self, line):
article_id = int(line.split(" ")[0])
if article_id in self.bad_articles:
return -1
if article_id not in self.article_ids:
try:
article = to_vector(line)
except IndexError:
self.bad_articles.add(article_id)
return -1
self.article_ids[article_id] = len(self.article_ids)
self.articles_1_d = np.append(self.articles_1_d, article).reshape(
[len(self.article_ids), self.d])
return article_id
def warmup(self, file):
self.warmup_impressions = 0
for line in file:
if self.warmup_impressions > self.training_size:
break
self.warmup_impressions += 1
line = line.split("|")
no_space_line = line[0].split(" ")
pre_selected_article = int(no_space_line[1])
click = int(no_space_line[2])
user = to_vector(line[1])
for article_line in line[2:]:
self.add_new_article(article_line)
# If preselected article is the one with bad format skip
if pre_selected_article in self.bad_articles:
continue
self.update(user, pre_selected_article, click)
def add_new_article(self, line):
article_id = int(line.split(" ")[0])
if article_id in self.bad_articles:
return -1
if article_id not in self.A:
try:
article = to_vector(line)
except IndexError:
print("Skipping line, weird formatting.." + str(article_id))
self.bad_articles.add(article_id)
return -1
self.articles[article_id] = article
self.A[article_id] = np.identity(self.d)
self.A_i[article_id] = np.identity(self.d)
self.B[article_id] = np.zeros(self.d * self.k).reshape(
[self.d, self.k])
self.b[article_id] = np.zeros(self.d)
return article_id
for rep in range(0, repetitions):
fo = open("..//..//R6//ydata-fp-td-clicks-v1_0.20090501", "r")
algo.warmup(fo)
print("Starting repetition {0}".format(rep))
total_impressions = 0.0
click_count = 0.0
impression_count = 0.0
local_clicks = 0.0
for line in fo:
total_impressions += 1
line = line.split("|")
no_space_line = line[0].split(" ")
pre_selected_article = int(no_space_line[1])
click = int(no_space_line[2])
user = to_vector(line[1])
selected_article, warmup = algo.select(user,
pre_selected_article,
line[2:],
total_impressions,
click)
# print(selected_article)
if selected_article == pre_selected_article and not warmup:
# print('.', end='', flush=True)
algo.update(user, pre_selected_article, click)
click_count += click
impression_count += 1
batch_size = 1000.0