def __init__(self, ranker, prels, anserini, index, topn=5, topw=3):
RelevanceFeedback.__init__(self,
ranker,
prels,
anserini,
index,
topn=topn)
self.topw = topw
def __init__(self, ranker, prels, anserini, index):
RelevanceFeedback.__init__(self,
ranker,
prels,
anserini,
index,
topn=10)
self.index_reader = pyserini.index.IndexReader(self.index)
def __init__(self, ranker, index, topn=10, topw=10, original_q_w=0.5):
RelevanceFeedback.__init__(self,
ranker=ranker,
prels=None,
anserini=None,
index=index,
topn=topn)
self.topw = topw
self.searcher = SimpleSearcher(index)
self.ranker = ranker
self.original_q_w = original_q_w
def get_rf_expanders(rankers, index, anserini, output):
expanders = []
for ranker in rankers:
ranker_name = get_ranker_name(ranker)
expanders.append(
RelevanceFeedback(ranker=ranker_name,
prels='{}.abstractqueryexpansion.{}.txt'.format(
output, ranker_name),
anserini=anserini,
index=index))
expanders.append(
Docluster(ranker=ranker_name,
prels='{}.abstractqueryexpansion.{}.txt'.format(
output, ranker_name),
anserini=anserini,
index=index)),
expanders.append(
Termluster(ranker=ranker_name,
prels='{}.abstractqueryexpansion.{}.txt'.format(
output, ranker_name),
anserini=anserini,
index=index))
expanders.append(
Conceptluster(ranker=ranker_name,
prels='{}.abstractqueryexpansion.{}.txt'.format(
output, ranker_name),
anserini=anserini,
index=index))
return expanders
def __init__(self,
ranker,
prels,
anserini,
index,
w_t,
w_a,
corpus_size,
replace=False,
topn=3,
topw=10,
adap=False):
RelevanceFeedback.__init__(self,
ranker,
prels,
anserini,
index,
topn=topn)
self.index_reader = pyserini.index.IndexReader(self.index)
self.topw = 10
self.adap = adap
self.w_t = w_t # weight for title field
self.w_a = w_a # weight for anchor field
self.corpus_size = corpus_size #total number of documents in the collection
def get_rf_expanders(rankers, corpus, output, ext_corpus=None):
expanders_list = []
for ranker in rankers:
ranker_name = utils.get_ranker_name(ranker)
if param.ReQue['expanders']['RM3']:
from expanders.rm3 import RM3
expanders_list.append(
RM3(ranker=ranker_name, index=param.corpora[corpus]['index']))
if param.ReQue['expanders']['RelevanceFeedback']:
from expanders.relevancefeedback import RelevanceFeedback
expanders_list.append(
RelevanceFeedback(
ranker=ranker_name,
prels='{}.abstractqueryexpansion.{}.txt'.format(
output, ranker_name),
anserini=param.anserini['path'],
index=param.corpora[corpus]['index']))
if param.ReQue['expanders']['Docluster']:
from expanders.docluster import Docluster
expanders_list.append(
Docluster(ranker=ranker_name,
prels='{}.abstractqueryexpansion.{}.txt'.format(
output, ranker_name),
anserini=param.anserini['path'],
index=param.corpora[corpus]['index'])),
if param.ReQue['expanders']['Termluster']:
from expanders.termluster import Termluster
expanders_list.append(
Termluster(ranker=ranker_name,
prels='{}.abstractqueryexpansion.{}.txt'.format(
output, ranker_name),
anserini=param.anserini['path'],
index=param.corpora[corpus]['index']))
if param.ReQue['expanders']['Conceptluster']:
from expanders.conceptluster import Conceptluster
expanders_list.append(
Conceptluster(ranker=ranker_name,
prels='{}.abstractqueryexpansion.{}.txt'.format(
output, ranker_name),
anserini=param.anserini['path'],
index=param.corpora[corpus]['index']))
if param.ReQue['expanders']['BertQE']:
from expanders.bertqe import BertQE
expanders_list.append(
BertQE(ranker=ranker_name,
prels='{}.abstractqueryexpansion.{}.txt'.format(
output, ranker_name),
index=param.corpora[corpus]['index'],
anserini=param.anserini['path']))
if param.ReQue['expanders']['OnFields']:
from expanders.onfields import OnFields
expanders_list.append(
OnFields(ranker=ranker_name,
prels='{}.abstractqueryexpansion.{}.txt'.format(
output, ranker_name),
anserini=param.anserini['path'],
index=param.corpora[corpus]['index'],
w_t=param.corpora[corpus]['w_t'],
w_a=param.corpora[corpus]['w_a'],
corpus_size=param.corpora[corpus]['size']))
if param.ReQue['expanders']['AdapOnFields']:
from expanders.adaponfields import AdapOnFields
expanders_list.append(
AdapOnFields(
ranker=ranker_name,
prels='{}.abstractqueryexpansion.{}.txt'.format(
output, ranker_name),
anserini=param.anserini['path'],
index=param.corpora[corpus]['index'],
w_t=param.corpora[corpus]['w_t'],
w_a=param.corpora[corpus]['w_a'],
corpus_size=param.corpora[corpus]['size'],
collection_tokens=param.corpora[corpus]['tokens'],
ext_corpus=ext_corpus,
ext_index=param.corpora[ext_corpus]['index'],
ext_collection_tokens=param.corpora[ext_corpus]['tokens'],
ext_w_t=param.corpora[ext_corpus]['w_t'],
ext_w_a=param.corpora[ext_corpus]['w_a'],
ext_corpus_size=param.corpora[ext_corpus]['size'],
adap=True))
return expanders_list