def change_domain_levels(src, dest, new_domain_level):
print "Processing", src
data = read_vm_file(src)
clicks = {}
for referrer, target, num_clicks in data:
if (
referrer is None
or target is None
or num_clicks == -1
or domain_level(referrer) == 1
or domain_level(target) == 1
):
continue
newr = change_domain_level(referrer, new_domain_level)
newt = change_domain_level(target, new_domain_level)
if newr not in clicks:
clicks[newr] = {}
if newt not in clicks[newr]:
clicks[newr][newt] = 0
clicks[newr][newt] += num_clicks
with open(dest, "w") as destf:
writer = csv.writer(destf, delimiter="\t")
for referrer in clicks:
for target in clicks[referrer]:
writer.writerow([referrer, target, clicks[referrer][target]])
def change_domain_levels(src, dest, new_domain_level):
print "Processing", src
data = read_vm_file(src)
clicks = {}
for referrer, target, num_clicks in data:
if referrer is None or target is None or num_clicks == -1\
or domain_level(referrer) == 1 or domain_level(target) == 1:
continue
newr = change_domain_level(referrer, new_domain_level)
newt = change_domain_level(target, new_domain_level)
if newr not in clicks:
clicks[newr] = {}
if newt not in clicks[newr]:
clicks[newr][newt] = 0
clicks[newr][newt] += num_clicks
with open(dest, 'w') as destf:
writer = csv.writer(destf, delimiter="\t")
for referrer in clicks:
for target in clicks[referrer]:
writer.writerow([referrer, target, clicks[referrer][target]])
compute_entropy(
os.path.join(os.getenv("TD"), "tweets",
"sample-u150-c100-news-only.txt"),
os.path.join(os.getenv("TR"), "twitter",
"twitter-avg-user-entropy-u150-c100-news-only.txt"),
os.path.join(os.getenv("TR"), "twitter",
"twitter-collective-entropy-u150-c100-news-only.txt"))
compute_entropy(
os.path.join(os.getenv("TD"), "aol", "sample-u150-c100-news-only.txt"),
os.path.join(os.getenv("TR"), "aol",
"aol-avg-user-entropy-u150-c100-news-only.txt"),
os.path.join(os.getenv("TR"), "aol",
"aol-collective-entropy-u150-c100-news-only.txt"))
compute_entropy(os.path.join(os.getenv("TD"), "tweets",
"sample-u1500-c1000.txt"),
os.path.join(os.getenv("TR"), "twitter",
"twitter-avg-user-entropy-u1500-c1000.txt"),
os.path.join(os.getenv("TR"), "twitter",
"twitter-collective-entropy-u1500-c1000.txt"),
url_mod=lambda url: change_domain_level(url, 2))
compute_entropy(os.path.join(os.getenv("TD"), "aol",
"sample-u1500-c1000.txt"),
os.path.join(os.getenv("TR"), "aol",
"aol-avg-user-entropy-u1500-c1000.txt"),
os.path.join(os.getenv("TR"), "aol",
"aol-collective-entropy-u1500-c1000.txt"),
url_mod=lambda url: change_domain_level(url, 2))
print "Writing result."
with open(collective_dest, 'w') as destf:
destf.write(str(h))
if __name__ == "__main__":
compute_entropy(
os.path.join(os.getenv("TD"), "tweets", "sample-u150-c100-news-only.txt"),
os.path.join(os.getenv("TR"), "twitter", "twitter-avg-user-entropy-u150-c100-news-only.txt"),
os.path.join(os.getenv("TR"), "twitter", "twitter-collective-entropy-u150-c100-news-only.txt")
)
compute_entropy(
os.path.join(os.getenv("TD"), "aol", "sample-u150-c100-news-only.txt"),
os.path.join(os.getenv("TR"), "aol", "aol-avg-user-entropy-u150-c100-news-only.txt"),
os.path.join(os.getenv("TR"), "aol", "aol-collective-entropy-u150-c100-news-only.txt")
)
compute_entropy(
os.path.join(os.getenv("TD"), "tweets", "sample-u1500-c1000.txt"),
os.path.join(os.getenv("TR"), "twitter", "twitter-avg-user-entropy-u1500-c1000.txt"),
os.path.join(os.getenv("TR"), "twitter", "twitter-collective-entropy-u1500-c1000.txt"),
url_mod = lambda url: change_domain_level(url, 2)
)
compute_entropy(
os.path.join(os.getenv("TD"), "aol", "sample-u1500-c1000.txt"),
os.path.join(os.getenv("TR"), "aol", "aol-avg-user-entropy-u1500-c1000.txt"),
os.path.join(os.getenv("TR"), "aol", "aol-collective-entropy-u1500-c1000.txt"),
url_mod = lambda url: change_domain_level(url, 2)
)
