def from_prebuilt_index(cls, prebuilt_index_name: str,
query_encoder: QueryEncoder):
"""Build a searcher from a pre-built index; download the index if necessary.
Parameters
----------
query_encoder: QueryEncoder
the query encoder, which has `encode` method that convert query text to embedding
prebuilt_index_name : str
Prebuilt index name.
Returns
-------
SimpleDenseSearcher
Searcher built from the prebuilt faiss index.
"""
print(
f'Attempting to initialize pre-built index {prebuilt_index_name}.')
try:
index_dir = download_prebuilt_index(prebuilt_index_name)
except ValueError as e:
print(str(e))
return None
print(f'Initializing {prebuilt_index_name}...')
return cls(index_dir, query_encoder)
def from_prebuilt_index(cls,
prebuilt_index_name: str,
query_encoder: Union[QueryEncoder, str],
min_idf=0):
"""Build a searcher from a pre-built index; download the index if necessary.
Parameters
----------
prebuilt_index_name : str
Prebuilt index name.
Returns
-------
LuceneSearcher
Searcher built from the prebuilt index.
"""
print(
f'Attempting to initialize pre-built index {prebuilt_index_name}.')
try:
index_dir = download_prebuilt_index(prebuilt_index_name)
except ValueError as e:
print(str(e))
return None
print(f'Initializing {prebuilt_index_name}...')
return cls(index_dir, query_encoder, min_idf)
def from_prebuilt_index(cls, prebuilt_index_name: str):
"""Build an index reader from the prebuilt index, download the index if necessary.
Parameters
----------
prebuilt_index_name : str
Prebuilt index name.
Returns
-------
IndexReader
Index reader built from the prebuilt index.
"""
index_dir = download_prebuilt_index(prebuilt_index_name)
return cls(index_dir)
def from_prebuilt_index(cls, prebuilt_index_name: str):
"""Build a searcher from the prebuilt index, download the index if necessary.
Parameters
----------
prebuilt_index_name : str
Prebuilt index name.
Returns
-------
SimpleSearcher
Searcher built from the prebuilt index.
"""
index_dir = download_prebuilt_index(prebuilt_index_name)
return cls(index_dir)
def test_bm25(self):
tmp_folder_name = self.tmp.split('/')[-1]
prebuilt_index_path = download_prebuilt_index('trec-covid-r3-abstract')
os.system(f'python {self.pyserini_root}/scripts/classifier_prf/rank_trec_covid.py \
-alpha 0.5 \
-clf lr \
-vectorizer tfidf \
-new_qrels {self.pyserini_root}/tools/topics-and-qrels/qrels.covid-round3.txt \
-base {self.tmp}/runs/covidex.t5.final.txt \
-tmp_base {tmp_folder_name} \
-qrels {self.pyserini_root}/tools/topics-and-qrels/qrels.covid-round2-cumulative.txt \
-index {prebuilt_index_path} \
-tag covidex.r3.t5.lr \
-output {self.tmp}/output.json')
with open(f'{self.tmp}/output.json') as json_file:
data = json.load(json_file)
self.assertEqual("0.3311\\n'", data['map'])
self.assertEqual("0.6866\\n'", data['ndcg'])
def from_prebuilt_index(cls, prebuilt_index_name: str):
"""Build a searcher from a pre-built index; download the index if necessary.
Parameters
----------
prebuilt_index_name : str
Prebuilt index name.
Returns
-------
SimpleSearcher
Searcher built from the prebuilt index.
"""
print(f'Attempting to initialize pre-built index {prebuilt_index_name}.')
try:
index_dir = download_prebuilt_index(prebuilt_index_name)
except ValueError as e:
print(str(e))
return None
print(f'Initializing {prebuilt_index_name}...')
return cls(index_dir)
def test_round5(self):
tmp_folder_name = self.tmp.split('/')[-1]
prebuilt_index_path = download_prebuilt_index('trec-covid-r5-abstract')
os.system(f'python {self.pyserini_root}/scripts/classifier_prf/rank_trec_covid.py \
-alpha 0.6 \
-clf lr \
-vectorizer tfidf \
-new_qrels {self.pyserini_root}/tools/topics-and-qrels/qrels.covid-round5.txt \
-base {self.tmp}/runs/covidex.r5.d2q.1s \
-tmp_base {tmp_folder_name} \
-qrels {self.pyserini_root}/tools/topics-and-qrels/qrels.covid-round4-cumulative.txt \
-index {prebuilt_index_path} \
-tag covidex.r5.d2q.1s \
-output {self.tmp}/output.json')
with open(f'{self.tmp}/output.json') as json_file:
data = json.load(json_file)
self.assertEqual("0.3859", data['map'])
self.assertEqual("0.8221", data['ndcg'])
os.system(f'python {self.pyserini_root}/scripts/classifier_prf/rank_trec_covid.py \
-alpha 0.6 \
-clf lr \
-vectorizer tfidf \
-new_qrels {self.pyserini_root}/tools/topics-and-qrels/qrels.covid-round5.txt \
-base {self.tmp}/runs/covidex.r5.d2q.2s \
-tmp_base {tmp_folder_name} \
-qrels {self.pyserini_root}/tools/topics-and-qrels/qrels.covid-round4-cumulative.txt \
-index {prebuilt_index_path} \
-tag covidex.r5.d2q.2s \
-output {self.tmp}/output.json')
with open(f'{self.tmp}/output.json') as json_file:
data = json.load(json_file)
self.assertEqual("0.3875", data['map'])
self.assertEqual("0.8304", data['ndcg'])
os.system(f'python {self.pyserini_root}/scripts/classifier_prf/rank_trec_covid.py \
-alpha 0.6 \
-clf lr \
-vectorizer tfidf \
-new_qrels {self.pyserini_root}/tools/topics-and-qrels/qrels.covid-round5.txt \
-base {self.tmp}/runs/covidex.r5.1s \
-tmp_base {tmp_folder_name} \
-qrels {self.pyserini_root}/tools/topics-and-qrels/qrels.covid-round4-cumulative.txt \
-index {prebuilt_index_path} \
-tag covidex.r5.1s \
-output {self.tmp}/output.json')
with open(f'{self.tmp}/output.json') as json_file:
data = json.load(json_file)
self.assertEqual("0.3885", data['map'])
self.assertEqual("0.8135", data['ndcg'])
os.system(f'python {self.pyserini_root}/scripts/classifier_prf/rank_trec_covid.py \
-alpha 0.6 \
-clf lr \
-vectorizer tfidf \
-new_qrels {self.pyserini_root}/tools/topics-and-qrels/qrels.covid-round5.txt \
-base {self.tmp}/runs/covidex.r5.2s \
-tmp_base {tmp_folder_name} \
-qrels {self.pyserini_root}/tools/topics-and-qrels/qrels.covid-round4-cumulative.txt \
-index {prebuilt_index_path} \
-tag covidex.r5.2s \
-output {self.tmp}/output.json')
with open(f'{self.tmp}/output.json') as json_file:
data = json.load(json_file)
self.assertEqual("0.3922", data['map'])
self.assertEqual("0.8311", data['ndcg'])
def from_prebuilt_index(cls, prebuilt_index_name: str):
index_dir = download_prebuilt_index(prebuilt_index_name)
return cls(index_dir)
