generate_rnr_feature_file.py 文件源码

def _print_feature_vectors_and_check_for_correct_answers(writer, rnr_search_results, qid, correct_ans_lookup):
    """
    write the search results to file as a feature vector with the qid and gt labels from the query.
    :param csv.writer writer:
    :param list(list(str)) rnr_search_results:
    :param str qid: the qid to print at the start of each feature vector
    :param dict(str,int) correct_ans_lookup: label lookup for correct answer ids
    :return: num_possible_correct, num_correct_answers_in_search_results
    :rtype: tuple(int,int)
    """
    num_possible_correct = len(correct_ans_lookup)
    num_correct_answers_in_search_results = 0

    for row in rnr_search_results:
        gt_label = 0
        doc_id = row[_ANS_ID_COL].strip()
        if doc_id in correct_ans_lookup:
            gt_label = correct_ans_lookup[doc_id]
            num_correct_answers_in_search_results += 1
        writer.writerow([qid] + row + [gt_label])

    return num_possible_correct, num_correct_answers_in_search_results