如何用Python及其库提取段落中带缩写的姓名？

我正在处理项目中的姓名提取任务，从Word文档提取的段落里，需要识别带缩写的完整姓名。段落格式有这些特点：

• 句首以Sri./SRI./Smt./SMT.开头，后面可能有空格或无空格
• 姓名中包含缩写（如M. S. ANANTHKRISHNA、GANGA A.K、SANTHOSH BHAT.N.C）
• 姓名后会出现S/o/D/o/W/o/C/o这类标识，或者直接跟年龄信息

之前尝试过正则匹配、NLTK Stanford NER、SpaCy的PROPN标注，都没能正确识别带缩写的姓名，现有代码的正则和字符串拆分方案效果不佳，求可行的Python实现。

示例段落：

Sri. SHASHANK KUMAR, S/o.Sri.Rajat Kumar, aged about 45 years, residing at ADDRESS Hereinafter called as the VENDOR.
Sri. M. S. ANANTHKRISHNA ,aged about 46 years, S/o. Sri. M.Neelakant residing at ADDRESSS. Hereinafter called the SECOND PARTY (PURCHASER).
Smt. GANGA A.K, D/o. Sri. Mukesh, aged about 31 years, residing at ADDRESS Hereinafter called the PURCHASER.
SRI.SANTHOSH BHAT.N.C S/o.Sri. Chikkabasappa, aged about 29 years, residing at ADDRESS Hereinafter called the PURCHASER.

现有相关代码：

def get_vendor_purchaser_details(df_hierchical_clustered_output):
    df_vendor_purchaser_details = df_hierchical_clustered_output[df_hierchical_clustered_output['k_means_labels_primary'] == 4][["id", "filename", "para_text"]]

    vendor_detail_list = []
    for index, row in df_vendor_purchaser_details.iterrows():
        vendor_detail = {}
        vendor_detail["id"] = row["id"]
        vendor_detail["para_text"] = row["para_text"]
        vendor_detail["name"] = get_vendor_name(row)

        vendor_detail_list.append(vendor_detail)
        print(vendor_detail["name"])

def get_vendor_name(entry):
    para_text = entry["para_text"]
    # 之前尝试的正则和拆分方案均无效
    # name_pattern = r'Sri. [A-Z\s]*|SRI.[A-Z\s]*|Smt. [A-z\s]*|SMT.[A-Z\s]*|SRI. [A-z\s]*|SMT. [A-Z\s]*'
    # vendor_name = re.findall(name_pattern,para_text)
    # ----------------------------------------------------
    # if "S/o" in para_text:
    #     vendor_name = para_text.split("S/o")[0]
    # elif "D/o" in para_text:
    #     vendor_name = para_text.split("D/o")[0]
    # elif "C/o" in para_text:
    #     vendor_name = para_text.split("C/o")[0]
    # elif "W/o" in para_text:
    #     vendor_name = para_text.split("W/o")[0]
    # else:
    #     vendor_name = ""
    # return vendor_name

针对你的段落格式，设计专门的正则表达式来匹配带缩写的姓名，核心思路：

1. 匹配开头的敬称（Sri./SRI./Smt./SMT.），处理后面有无空格的情况
2. 匹配姓名部分：允许大写字母、点号、空格，直到遇到S/o/D/o/W/o/C/o、逗号、年龄标识（aged）这类终止符
3. 清理匹配结果中的多余空格和标点

改进后的代码实现 #

import re

def get_vendor_name(entry):
    para_text = entry["para_text"]
    # 正则模式：匹配敬称开头，然后匹配姓名直到终止符
    name_pattern = r'(Sri\.|SRI\.|Smt\.|SMT\.)\s?([A-Z\s\.]+?)(?=\s*(S/o|D/o|W/o|C/o|aged|,))'
    matches = re.search(name_pattern, para_text, re.IGNORECASE)
    if matches:
        # 提取姓名部分，清理多余空格和末尾的点/空格
        full_name = matches.group(2).strip().rstrip('.')
        # 合并连续空格
        full_name = re.sub(r'\s+', ' ', full_name)
        return full_name
    return ""

# 测试示例
test_paragraphs = [
    "Sri. SHASHANK KUMAR, S/o.Sri.Rajat Kumar, aged about 45 years, residing at ADDRESS Hereinafter called as the VENDOR.",
    "Sri. M. S. ANANTHKRISHNA ,aged about 46 years, S/o. Sri. M.Neelakant residing at ADDRESSS. Hereinafter called the SECOND PARTY (PURCHASER).",
    "Smt. GANGA A.K, D/o. Sri. Mukesh, aged about 31 years, residing at ADDRESS Hereinafter called the PURCHASER.",
    "SRI.SANTHOSH BHAT.N.C S/o.Sri. Chikkabasappa, aged about 29 years, residing at ADDRESS Hereinafter called the PURCHASER."
]

for para in test_paragraphs:
    print(get_vendor_name({"para_text": para}))

测试结果 #

运行上述测试代码会输出：

SHASHANK KUMAR
M. S. ANANTHKRISHNA
GANGA A.K
SANTHOSH BHAT.N.C

正则说明 #

• (Sri\.|SRI\.|Smt\.|SMT\.)：匹配开头的敬称，转义点号避免匹配任意字符
• \s?：匹配敬称后可选的空格（处理有空格或无空格的情况）
• ([A-Z\s\.]+?)：非贪婪匹配姓名部分，允许大写字母、空格、点号（支持缩写）
• (?=\s*(S/o|D/o|W/o|C/o|aged|,))：正向预查终止符，确保匹配到姓名结束的位置，不包含终止符本身
• 后续的strip()、rstrip('.')和re.sub(r'\s+', ' ', full_name)用于清理多余的空格和末尾的点号，保证姓名格式整洁

内容的提问来源于stack exchange，提问作者Shashank Shandilya