semaj83/ctmatch
0
1 2from typing import List, Tuple3 4TREC_REL_PATH = "/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/trec_21_judgments.txt"5KZ_REL_PATH = "/Users/jameskelly/Documents/cp/ctmatch/data/kz_data/qrels-clinical_trials.txt"6 7TREC_RELLED_TOPIC_PATH = "/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/processed_trec_data/processed_trec21_topics.jsonl" 8KZ_RELLED_TOPIC_PATH = '/Users/jameskelly/Documents/cp/ctmatch/data/kz_data/processed_kz_data/processed_kz_topics.jsonl' 9 10KZ_DOC_PATH = '/Users/jameskelly/Documents/cp/ctmatch/data/kz_data/clinicaltrials.gov-16_dec_2015.zip'11KZ_PROCESSED_DOC_PATH = '/Users/jameskelly/Documents/cp/ctmatch/data/kz_data/processed_kz_data/processed_kz_docs.jsonl'12 13TREC_ML_PATH = '/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/trec_data.jsonl'14KZ_ML_PATH = '/Users/jameskelly/Documents/cp/ctmatch/data/kz_data/kz_data.jsonl'15 16 17def get_data_tuples(trec_or_kz: str = 'trec') -> List[Tuple[str, str]]:18 if trec_or_kz == 'trec':19 return get_trec_doc_data_tuples(), get_trec_topic_data_tuples()20 return get_kz_doc_data_tuples(), get_kz_topic_data_tuples()21 22 23 24# --------------------------------------------------------------------------------------------------------------- #25# data from TREC clinical track 2021 & 202226# --------------------------------------------------------------------------------------------------------------- #27 28 29def get_trec_doc_data_tuples() -> List[Tuple[str]]:30 trec22_pt1_docs = '/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/trec_docs_21/ClinicalTrials.2021-04-27.part1.zip'31 trec_pt1_target = '/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/processed_trec_data/processed_trec22_docs_part1.jsonl'32 33 trec22_pt2_docs = '/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/trec_docs_21/ClinicalTrials.2021-04-27.part2.zip'34 trec_pt2_target = '/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/processed_trec_data/processed_trec22_docs_part2.jsonl'35 36 trec22_pt3_docs = '/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/trec_docs_21/ClinicalTrials.2021-04-27.part3.zip'37 trec_pt3_target = '/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/processed_trec_data/processed_trec22_docs_part3.jsonl'38 39 trec22_pt4_docs = '/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/trec_docs_21/ClinicalTrials.2021-04-27.part4.zip'40 trec_pt4_target = '/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/processed_trec_data/processed_trec22_docs_part4.jsonl'41 42 trec22_pt5_docs = '/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/trec_docs_21/ClinicalTrials.2021-04-27.part5.zip'43 trec_pt5_target = '/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/processed_trec_data/processed_trec22_docs_part5.jsonl'44 45 trec_doc_data_tuples = [46 (trec22_pt1_docs, trec_pt1_target), 47 (trec22_pt2_docs, trec_pt2_target), 48 (trec22_pt3_docs, trec_pt3_target),49 (trec22_pt4_docs, trec_pt4_target),50 (trec22_pt5_docs, trec_pt5_target)51 ]52 53 return trec_doc_data_tuples54 55 56def get_trec_topic_data_tuples() -> List[Tuple[str]]:57 trec21_topic_path = '/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/trec_21_topics.xml'58 trec21_topic_target = '/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/processed_trec_data/processed_trec21_topics.jsonl'59 trec22_topic_path = '/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/trec_22_topics.xml'60 trec22_topic_target = '/Users/jameskelly/Documents/cp/ctmatch/data/trec_data/processed_trec_data/processed_trec22_topics.jsonl'61 62 trec_topic_data_tuples = [63 (trec21_topic_path, trec21_topic_target), 64 (trec22_topic_path, trec22_topic_target)65 ]66 return trec_topic_data_tuples67 68 69 70 71# --------------------------------------------------------------------------------------------------------------- #72# data from Koontz, et al. (2016)73# --------------------------------------------------------------------------------------------------------------- #74def get_kz_doc_data_tuples() -> List[Tuple[str]]:75 # kz_doc_data_tuples = []76 # for i in range(1, 18):77 # kz_doc_path = f'/Users/jameskelly/Documents/cp/ctmatch/data/kz_data/kz_doc_splits/kz_doc_split{i}.zip'78 # kz_doc_target = f'/Users/jameskelly/Documents/cp/ctmatch/data/kz_data/processed_kz_data/processed_kz_doc_split{i}.jsonl'79 # kz_doc_data_tuples.append((kz_doc_path, kz_doc_target))80 kz_docs = KZ_DOC_PATH81 kz_docs_target = KZ_PROCESSED_DOC_PATH82 return [(kz_docs, kz_docs_target)]83 84 #return kz_doc_data_tuples85 86def get_kz_topic_data_tuples() -> List[Tuple[str]]:87 kz_topic_desc_path = '/Users/jameskelly/Documents/cp/ctmatch/data/kz_data/topics-2014_2015-description.topics'88 kz_topic_target = '/Users/jameskelly/Documents/cp/ctmatch/data/kz_data/processed_kz_data/processed_kz_topics.jsonl'89 kz_topic_data_tuples = [90 (kz_topic_desc_path, kz_topic_target)91 ]92 return kz_topic_data_tuples93 