def gen_entity_net_20ng(): proper_entity_dict_file = 'e:/dc/20ng_bydate/entity_names.txt' doc_entity_candidates_file = 'e:/dc/20ng_bydate/doc_entity_candidates.txt' dst_doc_entity_list_file = 'e:/dc/20ng_bydate/doc_entities_short.bin' # gen_doc_entity_list(proper_entity_dict_file, doc_entity_candidates_file, dst_doc_entity_list_file) dst_entity_cnts_file = 'e:/dc/20ng_bydate/entity_cnts.bin' text_process_common.gen_word_cnts_file_from_bow_file(dst_doc_entity_list_file, dst_entity_cnts_file) entity_candidate_cliques_file = 'e:/dc/20ng_bydate/entity_candidate_cliques.txt' dst_entity_net_adj_list_file = 'e:/dc/20ng_bydate/entity_net_adj_list.bin'
def gen_entity_net_wiki(): proper_entity_dict_file = 'e:/dc/el/wiki/entity_names.txt' doc_entity_candidates_file = 'e:/dc/el/wiki/doc_entity_candidates.txt' dst_doc_entity_list_file = 'e:/dc/el/wiki/wiki_entities.bin' gen_doc_entity_pairs(proper_entity_dict_file, doc_entity_candidates_file, dst_doc_entity_list_file) dst_entity_cnts_file = 'e:/dc/el/wiki/entity_cnts.bin' text_process_common.gen_word_cnts_file_from_bow_file(dst_doc_entity_list_file, dst_entity_cnts_file) entity_candidate_cliques_file = 'e:/dc/el/wiki/entity_candidate_cliques.txt' dst_entity_net_adj_list_file = 'e:/dc/el/wiki/entity_net_adj_list.bin' gen_entity_entity_pairs(proper_entity_dict_file, entity_candidate_cliques_file, dst_entity_net_adj_list_file)