我的方法是使用第三个包，它支持负向后查找的无限宽度，以忽略所有特殊情况
regex包可能会有所帮助

(?<!U\.S|Ser|No|Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec|Pat|Fig)\.(?= [A-Z\d])

演示：regex101（注意，我在这里使用.NET引擎而不是Python，因为它支持无限lookbehind进行演示）

说明：

• (?<!U\.S|Ser|No|Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec|Pat|Fig)负向后查找，忽略所有以这些特殊字开头的.。
• \.匹配点.
• (?= [A-Z\d])：向前看，确保后面有空格和大写字符或数字

安装regex包

pip instal regex

编程：

import regex
s = "This Application is a Continuation of U.S. patent application Ser. No. 15/731,069, filed on Apr. 14, 2017, which was a Continuation of U.S. application Ser. No. 14/998,574, filed on Jan. 21, 2016 which was a Continuation of U.S. application Ser. No. 14/198,695 (now U.S. Pat. No. 9,286,457) filed on Mar. 6, 2014, which was a Continuation in Part of U.S. patent application Ser. No. 12/931,340, filed on Jan. 31, 2011, (now U.S. Pat. No. 8,842,887) which was a Continuation in Part of U.S. patent application Ser. No. 12/627,413 filed on Nov. 30, 2009, (now U.S. Pat. No. 7,916,907) which was a continuation of application Ser. No. 11/151,412, filed on Jun. 14, 2005, now abandoned. application Ser. No. 14/198,695 claims the benefit of Provisional Application No. 61/851,884, filed on Mar. 15, 2013. Application Ser. No. 12/931,340 claims the benefit of Provisional Application No. 61/456,901, filed on Nov. 15, 2010 and application Ser. No. 11/151,412 claims the benefit of Provisional Application No. 60/579,422 filed on Jun. 14, 2004. Testing some more example of U.S.A and U.S in my paragraph. Checking Fig. 3. in my paragraph. 1 new sentences added to this text block."
lines = regex.split(r"(?<!U\.S|Ser|No|Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec|Pat|Fig)\.(?= [A-Z\d])", s)
for line in lines:
  print(line)

输出：

This Application is a Continuation of U.S. patent application Ser. No. 15/731,069, filed on Apr. 14, 2017, which was a Continuation of U.S. application Ser. No. 14/998,574, filed on Jan. 21, 2016 which was a Continuation of U.S. application Ser. No. 14/198,695 (now U.S. Pat. No. 9,286,457) filed on Mar. 6, 2014, which was a Continuation in Part of U.S. patent application Ser. No. 12/931,340, filed on Jan. 31, 2011, (now U.S. Pat. No. 8,842,887) which was a Continuation in Part of U.S. patent application Ser. No. 12/627,413 filed on Nov. 30, 2009, (now U.S. Pat. No. 7,916,907) which was a continuation of application Ser. No. 11/151,412, filed on Jun. 14, 2005, now abandoned. application Ser. No. 14/198,695 claims the benefit of Provisional Application No. 61/851,884, filed on Mar. 15, 2013
 Application Ser. No. 12/931,340 claims the benefit of Provisional Application No. 61/456,901, filed on Nov. 15, 2010 and application Ser. No. 11/151,412 claims the benefit of Provisional Application No. 60/579,422 filed on Jun. 14, 2004
 Testing some more example of U.S.A and U.S in my paragraph
 Checking Fig. 3. in my paragraph
 1 new sentences added to this text block.

**注意：**如果没有像我最后添加的句子1 new sentences added to this text block.那样以数字开头的行，则可以使用更简单的版本：

(?<!U\.S|Ser|No|Pat|Fig)\.(?= [A-Z])

• 编辑：因为在这种情况下，正则表达式可能会变得非常复杂，所以我会选择使用库。
  使用节库

如果不需要正则表达式，一个选择是使用stanza：

import stanza

stanza.download('en')
# define pipeline
nlp = stanza.Pipeline(lang="en", processors='tokenize')
# create document
doc = nlp(text)
# extract sentences
doc_sentences = [sentence.text for sentence in doc.sentences]

for s in doc_sentences:
    print(s)

这给出了以下结果，即使在句子末尾使用缩写也能很好地工作（原始正则表达式无法处理）：

This Application is a Continuation of U.S. patent application Ser. No. 15/731,069, filed on Apr. 14, 2017, which was a Continuation of U.S. application Ser. No. 14/998,574, filed on Jan. 21, 2016 which was a Continuation of U.S. application Ser. No. 14/198,695 (now U.S. Pat. No. 9,286,457) filed on Mar. 6, 2014, which was a Continuation in Part of U.S. patent application Ser. No. 12/931,340, filed on Jan. 31, 2011, (now U.S. Pat. No. 8,842,887) which was a Continuation in Part of U.S. patent application Ser. No. 12/627,413 filed on Nov. 30, 2009, (now U.S. Pat. No. 7,916,907) which was a continuation of application Ser. No. 11/151,412, filed on Jun. 14, 2005, now abandoned.
application Ser. No. 14/198,695 claims the benefit of Provisional Application No. 61/851,884, filed on Mar. 15, 2013.
Application Ser. No. 12/931,340 claims the benefit of Provisional Application No. 61/456,901, filed on Nov. 15, 2010 and application Ser. No. 11/151,412 claims the benefit of Provisional Application No. 60/579,422 filed on Jun. 14, 2004.
Testing some more example of U.S.A and U.S in my paragraph.
Checking Fig. 3. in my paragraph about the U.S.
The check looks good.

• 注意：为了测试这一点，我稍微修改了你的输入文本，在句子的结尾有一个缩写，我添加了：Checking Fig. 3. in my paragraph about the U.S. The check looks good.*

如果正则表达式对您来说不是必需的，我建议您测试一下，因为用正则表达式覆盖所有边缘情况可能会很麻烦。

原始正则表达式方法

一种方法是更好地定义/简化“句子的结尾”是什么，以避免需要定义所有特殊情况。在您的示例中，这似乎是一个合理的简化：

• 只有在以下情况下，句子才会结束：一个点后跟一个空格，后面跟着一个以大写字母开头并且至少有4个字符长的单词。*

使用这种简化，您可以使用lookaheadAssert来匹配所有出现的“the end of a sentence”\.\s(?=[A-Z][a-zA-Z]{3,})，并使用此表达式来拆分使用re.split提供的文本，如下所示：

import re

text = "<your text>"

sentences = re.split(r"\.\s(?=[A-Z][a-zA-Z]{3,})", text)

print(sentences)

根据你的数据，这些是用这种方法得到的句子：

This Application is a Continuation of U.S. patent application Ser. No. 15/731,069, filed on Apr. 14, 2017, which was a Continuation of U.S. application Ser. No. 14/998,574, filed on Jan. 21, 2016 which was a Continuation of U.S. application Ser. No. 14/198,695 (now U.S. Pat. No. 9,286,457) filed on Mar. 6, 2014, which was a Continuation in Part of U.S. patent application Ser. No. 12/931,340, filed on Jan. 31, 2011, (now U.S. Pat. No. 8,842,887) which was a Continuation in Part of U.S. patent application Ser. No. 12/627,413 filed on Nov. 30, 2009, (now U.S. Pat. No. 7,916,907) which was a continuation of application Ser. No. 11/151,412, filed on Jun. 14, 2005, now abandoned. application Ser. No. 14/198,695 claims the benefit of Provisional Application No. 61/851,884, filed on Mar. 15, 2013
Application Ser. No. 12/931,340 claims the benefit of Provisional Application No. 61/456,901, filed on Nov. 15, 2010 and application Ser. No. 11/151,412 claims the benefit of Provisional Application No. 60/579,422 filed on Jun. 14, 2004
Testing some more example of U.S.A and U.S in my paragraph
Checking Fig. 3. in my paragraph.

根据您希望分割位置的具体程度，您可能会使用re.sub的交替
然后在回调中检查模式中是否存在捕获组。如果捕获组存在，则在替换中使用它，后面跟着2个换行符（或者您想在它后面放置的内容）
在另一种情况下，只有匹配x.group()，它应该留在那里，所以你可以把它放在替换中。

(?<!\S)[A-Z](?:[a-z]+\.(?:\s+\d+\.)?|(?:\.[A-Z]+)*\.?)(?!\S)|([!.?])\s(?=[A-Z])

图案吻合

• (?<!\S)向左声明空白边界
• [A-Z]匹配字符A-Z
• (?:备选项的非捕获组
• [a-z]+\.匹配1+字符a-z和一个.
• (?:\s+\d+\.)?可选匹配1+空格字符、1+数字和.
• |或
• (?:\.[A-Z]+)*\.?匹配.的可选重复，后跟A-Z和可选点
• )关闭非捕获组
• (?!\S)在右侧声明空白边界
• |或
• ([!.?])\s捕获group 1，捕获!.?中的一个并匹配空白字符
• (?=[A-Z])正向预测，直接在右侧Assert字符A-Z

请参见regex demo和Python demo

import re

pattern = r"(?<!\S)[A-Z](?:[a-z]+\.(?:\s+\d+\.)?|(?:\.[A-Z]+)*\.?)(?!\S)|([!.?])\s(?=[A-Z])"
s = "This Application is a Continuation of U.S. patent application Ser. No. 15/731,069, filed on Apr. 14, 2017, which was a Continuation of U.S. application Ser. No. 14/998,574, filed on Jan. 21, 2016 which was a Continuation of U.S. application Ser. No. 14/198,695 (now U.S. Pat. No. 9,286,457) filed on Mar. 6, 2014, which was a Continuation in Part of U.S. patent application Ser. No. 12/931,340, filed on Jan. 31, 2011, (now U.S. Pat. No. 8,842,887) which was a Continuation in Part of U.S. patent application Ser. No. 12/627,413 filed on Nov. 30, 2009, (now U.S. Pat. No. 7,916,907) which was a continuation of application Ser. No. 11/151,412, filed on Jun. 14, 2005, now abandoned. application Ser. No. 14/198,695 claims the benefit of Provisional Application No. 61/851,884, filed on Mar. 15, 2013. Application Ser. No. 12/931,340 claims the benefit of Provisional Application No. 61/456,901, filed on Nov. 15, 2010 and application Ser. No. 11/151,412 claims the benefit of Provisional Application No. 60/579,422 filed on Jun. 14, 2004. Testing some more example of U.S.A and U.S in my paragraph. Checking Fig. 3. in my paragraph."

result = re.sub(pattern, lambda x: x.group(1) + "\n\n" if x.group(1) else x.group(), s)
print(result)

输出

This Application is a Continuation of U.S. patent application Ser. No. 15/731,069, filed on Apr. 14, 2017, which was a Continuation of U.S. application Ser. No. 14/998,574, filed on Jan. 21, 2016 which was a Continuation of U.S. application Ser. No. 14/198,695 (now U.S. Pat. No. 9,286,457) filed on Mar. 6, 2014, which was a Continuation in Part of U.S. patent application Ser. No. 12/931,340, filed on Jan. 31, 2011, (now U.S. Pat. No. 8,842,887) which was a Continuation in Part of U.S. patent application Ser. No. 12/627,413 filed on Nov. 30, 2009, (now U.S. Pat. No. 7,916,907) which was a continuation of application Ser. No. 11/151,412, filed on Jun. 14, 2005, now abandoned. application Ser. No. 14/198,695 claims the benefit of Provisional Application No. 61/851,884, filed on Mar. 15, 2013.

Application Ser. No. 12/931,340 claims the benefit of Provisional Application No. 61/456,901, filed on Nov. 15, 2010 and application Ser. No. 11/151,412 claims the benefit of Provisional Application No. 60/579,422 filed on Jun. 14, 2004.

Testing some more example of U.S.A and U.S in my paragraph.

Checking Fig. 3. in my paragraph.