Deeplearning/test_another_dataset_metrics.py at main · FakeLearne/Deeplearning · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
#   BIO标签抽取实体
def extract_entities(sentence, bio):
    words = list(sentence)  # 句子按字拆分
    bio_tags = bio.split()  # BIO 标签按空格拆分
    if len(words)>len(bio_tags):
        words = words[:len(bio_tags)]
    assert len(words) == len(bio_tags), f"句子长度和BIO标签长度不匹配！{len(sentence)},{len(bio)},{words},{bio_tags}"
    entities = []
    current_entity = []
    current_label = None

    for char, tag in zip(words, bio_tags):
        if tag.startswith("B-"):  # 遇到新的实体
            if current_entity:  # 先存储上一个实体
                entities.append("".join(current_entity))
            current_entity = [char]  # 记录当前字符
            current_label = tag[2:]  # 提取实体类别（去掉B-前缀）

        elif tag.startswith("I-") and current_label == tag[2:]:  # 继续当前实体
            current_entity.append(char)

        else:  # 遇到 "O" 或者 I- 但不匹配当前实体
            if current_entity:
                entities.append("".join(current_entity))
                current_entity = []
                current_label = None

    if current_entity:  # 处理最后一个实体
        entities.append("".join(current_entity))

    return entities


#   计算recall,precision,f1分数
def calculate_metrics(true_labels, predicted_labels):
    # 计算TP, FP, FN
    true_set = set(true_labels)
    predicted_set = set(predicted_labels)

    # 计算交集、并集
    true_positive = len(true_set & predicted_set)  # 交集
    false_positive = len(predicted_set - true_set)  # 预测正确但实际没有的实体
    false_negative = len(true_set - predicted_set)  # 实际正确但未预测出的实体

    # 计算precision, recall, f1
    precision = true_positive / (true_positive + false_positive) if true_positive + false_positive > 0 else 0
    recall = true_positive / (true_positive + false_negative) if true_positive + false_negative > 0 else 0
    f1 = 2 * precision * recall / (precision + recall) if precision + recall > 0 else 0

    return precision, recall, f1


if __name__=="__main__":

    # predict_file = 'D:/PyProject/bert_github/BERT-for-Sequence-Labeling-and-Text-Classification-master/output_model/ChemNER+_bertcrf_epoch50/test_predictions.txt'
    predict_file = 'D:/PyProject/bert_github/BERT-for-Sequence-Labeling-and-Text-Classification-master/result/ours_roberta_on_ChemNER+_predictions.txt'

    sentence_file = 'ChemNER+/test/seq.in'
    label_file = 'ChemNER+/test/seq.out'
    sentences = []
    real_bios = []
    predict_bios = []

    with open(sentence_file, 'r', encoding='utf-8') as f1:   #句子
        lines = f1.readlines()
        for line in lines:
            sentences.append(line)
    with open(label_file, 'r', encoding='utf-8') as f2:      #真实标签
        lines = f2.readlines()
        for line in lines:
            real_bios.append(line)
    with open(predict_file, 'r', encoding='utf-8') as f3:    #预测结果
        lines = f3.readlines()
        for i,line in enumerate(lines):
            tokens = line.strip().split()
            tokens = [t for t in tokens if t not in ["[CLS]","[SEP]"]]

            predict_bios.append(" ".join(tokens))
            # print(predict_bios[i])
    real = []
    predict = []
    precisions, recalls,f1s = [], [], []
    for i in range(len(sentences)):
        sentence = sentences[i]
        real_bio = real_bios[i]
        predict_bio = predict_bios[i]
        real_entities = extract_entities(sentence, real_bio)
        predict_enities = extract_entities(sentence, predict_bio)
        # print(real_entities)
        # print(predict_enities)
        #   跳过空标签
        if not real_entities:
            # print(real_entities)
            continue

        precision, recall, f1 = calculate_metrics(real_entities, predict_enities)
        #   记录结果
        precisions.append(precision)
        recalls.append(recall)
        f1s.append(f1)

        real.append(real_entities)
        predict.append(predict_enities)

    avg_precision = sum(precisions) / len(precisions) if precisions else 0
    avg_recall = sum(recalls) / len(recalls) if recalls else 0
    avg_f1 = sum(f1s) / len(f1s) if f1s else 0

    print(f"Average Precision: {avg_precision:.4f}")
    print(f"Average Recall: {avg_recall:.4f}")
    print(f"Average F1 Score: {avg_f1:.4f}")