def run_for_question(question, number):
q = Question(number, question, '../docs/top_docs.%d.gz'%number)
query = BuildQuery.buildQuery(question)
documents = q.search(query)
rankings = q.golden_passage_retriever(documents)
passage_metrics = q.test(rankings)
return map(lambda x: float(sum(x)) / len(x), zip(*passage_metrics))
def justDoIR():
questions = extract_questions(questions_file)
output = open("../IRoutput.txt", 'w')
for qNumber in questions:
docs = DOCS + "top_docs." + str(qNumber) + ".gz"
question = Question(qNumber,questions[qNumber],docs)
# Can do baseline here or do full process.
# Get expanded question:
query = questions[qNumber]
query = BuildQuery.buildFullQuery(query, hyponyms=True, hypernyms=True)
ir_results = question.search(query)
for (_,answer) in ir_results:
output.write(answer+"\n")
output.close()
def main():
questions = extract_questions(questions_file)
output = open(output_file,"w")
for qNumber in questions:
print "\n"
docs = DOCS + "top_docs." + str(qNumber) + ".gz"
#print questions[qNumber]
question = Question(qNumber,questions[qNumber],docs)
# Can do baseline here or do full process.
# Get expanded question:
query = BuildQuery.buildFullQuery(questions[qNumber], hyponyms=False, hypernyms=False)
print query
ir_results = question.search(query)
#print ir_results
passages = question.golden_passage_retriever(ir_results)
top5 = question.top5(passages)
for answer in top5:
print answer
output.write(str(qNumber) + " top_docs." + str(qNumber) + " " + answer + "\n")
output.close()
if not IS_TEST:
score = MRR()
print score
print "done"
#! /usr/bin/python
from Question import Question
import BuildQuery
question = 'Where is Belize located?'
number = 202
def run_for_question(question, number):
q = Question(number, question, '../docs/top_docs.%d.gz'%number)
query = BuildQuery.buildQuery(question)
documents = q.search(query)
rankings = q.golden_passage_retriever(documents)
passage_metrics = q.test(rankings)
return map(lambda x: float(sum(x)) / len(x), zip(*passage_metrics))
questions = BuildQuery.getQuestions()
ranking_list = []
base = 201
for i in range(0,15):
ranking_list.append(run_for_question(questions[i][1], questions[i][0]))
avg = map(lambda x: float(sum(x)) / len(x), zip(*ranking_list))
print avg
