def synthesize(client, featureSet='release', latestTrainingSetOut='data/latest-training.txt', classicTrainingSetOut='data/classic-training.txt'):
from ltr.judgments import judgments_to_file, Judgment
print('Generating ratings for classic and latest model')
NO_ZERO = False
resp = client.log_query('tmdb', 'release', None)
# Classic film fan
judgments = []
print("Generating Classic judgments:")
for hit in resp:
rating = get_classic_rating(hit['ltr_features'][0])
if rating == 0 and NO_ZERO:
continue
judgments.append(Judgment(qid=1,docId=hit['id'],grade=rating,features=hit['ltr_features'],keywords=''))
with open(classicTrainingSetOut, 'w') as out:
judgments_to_file(out, judgments)
# Latest film fan
judgments = []
print("Generating Recent judgments:")
for hit in resp:
rating = get_latest_rating(hit['ltr_features'][0])
if rating == 0 and NO_ZERO:
continue
judgments.append(Judgment(qid=1,docId=hit['id'],grade=rating,features=hit['ltr_features'],keywords=''))
with open(latestTrainingSetOut, 'w') as out:
judgments_to_file(out, judgments)
def synthesize(client,
featureSet='release',
latestTrainingSetOut='data/latest-training.txt',
classicTrainingSetOut='data/classic-training.txt'):
from ltr.judgments import judgments_to_file, Judgment
print('Generating ratings for classic and latest model')
NO_ZERO = False
resp = client.log_query('tmdb', 'release', None)
docs = []
for hit in resp:
# Expect clients to return features per doc in ltr_features as ordered list
feature = hit['ltr_features'][0]
docs.append([feature]) # Treat features as ordered lists
# Classic film fan
judgments = []
for fv in docs:
rating = get_classic_rating(fv[0])
if rating == 0 and NO_ZERO:
continue
judgments.append(
Judgment(qid=1,
docId=rating,
grade=rating,
features=fv,
keywords=''))
with open(classicTrainingSetOut, 'w') as out:
judgments_to_file(out, judgments)
judgments = []
for fv in docs:
rating = get_latest_rating(fv[0])
if rating == 0 and NO_ZERO:
continue
judgments.append(
Judgment(qid=1,
docId=rating,
grade=rating,
features=fv,
keywords=''))
with open(latestTrainingSetOut, 'w') as out:
judgments_to_file(out, judgments)
print('Done')
def rate_results():
from sys import argv
esUrl = 'http://localhost:9200/tmdb/'
judgFile = argv[1]
full_judgments, existing_kws, last_qid = load_judgments(judgFile)
keywords = "-"
new_qid = last_qid + 1
while len(keywords) > 0:
input_kws = input("Enter the Keywords ('GTFO' to exit) ")
if input_kws == "GTFO":
break
keywords, search_with, orig_query_judgments, existing_qid, fuzzy, copy_src_kws =\
handleKeywords(input_kws, existing_kws, full_judgments)
curr_qid = 0
if existing_qid > 0:
curr_qid = existing_qid
print("Updating judgments for qid:%s" % curr_qid)
else:
existing_kws.add(keywords)
curr_qid = new_qid
print("New Keywords %s qid:%s" % (keywords, curr_qid))
new_qid += 1
new_query_judgments = []
if copy_src_kws is not None:
print("Copying from %s <- %s" % (keywords, copy_src_kws))
for judg in orig_query_judgments:
judgment = Judgment(int(judg.grade),
qid=new_qid,
keywords=keywords,
docId=judg.docId)
new_query_judgments.append(judgment)
existing_kws.add(keywords)
curr_qid = new_qid
new_qid += 1
else:
results = get_potential_results(esUrl, search_with, fuzzy)
new_query_judgments = grade_results(results, keywords, curr_qid)
foldInNewRatings(full_judgments, orig_query_judgments,
new_query_judgments)
with open(judgFile, 'w') as f:
judgments_to_file(f, full_judgments)
judgFile = argv[1]
fullJudgments, existingKws, lastQid = loadJudgments(judgFile)
keywords = "-"
newQid = lastQid + 1
while len(keywords) > 0:
inputKws = input("Enter the Keywords ('GTFO' to exit) ")
if inputKws == "GTFO":
break
keywords, searchWith, origQueryJudgments, existingQid, fuzzy = handleKeywords(
inputKws, fullJudgments)
currQid = 0
if existingQid > 0:
currQid = existingQid
print("Updating judgments for qid:%s" % currQid)
else:
existingKws.add(keywords)
currQid = newQid
print("New Keywords %s qid:%s" % (keywords, currQid))
newQid += 1
results = getPotentialResults(esUrl, searchWith, fuzzy)
newQueryJudgments = gradeResults(results, keywords, currQid)
foldInNewRatings(fullJudgments, origQueryJudgments, newQueryJudgments)
judgments_to_file(judgFile, fullJudgments)