如何使用Python库spaCy检测句子的主动与被动语态？

要判断英文句子的语态，核心是利用spaCy的依存句法分析功能：

• 主动语态的核心谓语动词，其主语的依存关系标记为nsubj（普通主语）
• 被动语态的核心谓语动词，其主语的依存关系标记为nsubjpass（被动主语）

具体实现步骤 #

1. 先安装spaCy和英文模型

pip install spacy
python -m spacy download en_core_web_sm

2. 编写检测函数，遍历句子中的动词判断语态

import spacy

nlp = spacy.load("en_core_web_sm")

def detect_voice(sentence):
    doc = nlp(sentence)
    for token in doc:
        # 检查被动主语标记或被动助动词（如was/were）
        if token.dep_ == "nsubjpass" or token.dep_ == "auxpass":
            return "被动语态"
    return "主动语态"

# 测试示例句子
passive_sentence = "John was accused of committing crimes by David"
active_sentence = "David accused John of committing crimes"

print(detect_voice(passive_sentence))  # 输出：被动语态
print(detect_voice(active_sentence))   # 输出：主动语态

提取核心语态结构（对应示例需求） #

如果需要像示例那样提取核心的语态结构，可以扩展函数：

def extract_core_voice_structure(sentence):
    doc = nlp(sentence)
    for token in doc:
        if token.dep_ == "nsubjpass":
            # 提取被动结构：主语 + 被动助动词 + 核心动词
            core_parts = [token.text]
            root_verb = token.head
            # 查找被动助动词（如was）
            for child in root_verb.children:
                if child.dep_ == "auxpass":
                    core_parts.append(child.text)
            core_parts.append(root_verb.text)
            return " ".join(core_parts)
        elif token.dep_ == "nsubj":
            # 提取主动结构：主语 + 核心动词
            return f"{token.text} {token.head.text}"
    return ""

print(extract_core_voice_structure(passive_sentence))  # 输出：John was accused
print(extract_core_voice_structure(active_sentence))   # 输出：David accused John

内容的提问来源于stack exchange，提问作者sruthi