Medspacy 无法检测到任何实体，尽管它存在

Question

我正在处理包含生物医学实体的文本。然而 medspacy 软件包未能检测到这些：

import medspacy

nlp = medspacy.load()    
text = "The patient was treated with warfarin for atrial fibrillation."
doc = nlp(text)
for ent in doc.ents:
    print(f"Entity: {ent.text}, Label: {ent.label_}")
print("Total Entities: " + str(len(doc.ents)))

输出： 实体总数：0

我正在尝试检查是否存在用于实体检测的任何管道：

print(nlp.pipe_names)

，我可以布局管道：

['medspacy_pyrush', 'medspacy_target_matcher', 'medspacy_context']

。你能帮我找到这个问题的解决方案吗？谢谢！

Answer 1

您还需要管道的

TargetMatcher

来检测实体。这是代码片段：

import medspacy
from medspacy.ner import TargetMatcher, TargetRule

# Load medspacy pipeline
nlp = medspacy.load()

# Initialize a TargetMatcher
target_matcher = nlp.get_pipe("medspacy_target_matcher")

# Create TargetRule objects for entity patterns
target_rules = [
    TargetRule("warfarin", "MEDICATION"),
    TargetRule("atrial fibrillation", "CONDITION")
]

target_matcher.add(target_rules) 

其余代码保持不变：

text = "The patient was treated with warfarin for atrial fibrillation."
doc = nlp(text)
for ent in doc.ents:
    print(f"Entity: {ent.text}, Label: {ent.label_}")
print("Total Entities: " + str(len(doc.ents)))

输出：

Entity: warfarin, Label: MEDICATION
Entity: atrial fibrillation, Label: CONDITION
Total Entities: 2