可能的解决方案：

out = df['col'].str.replace(':', r':\User *').str.split(r':').explode()
(out[~(out.str.contains('Hello|Hi') | out.eq('User '))].str.replace(r'*', ':')
 .groupby(level=0).sum()
 .str.replace(r' \User :', ':'))

输出：

0    \User : Order status.\nBot: Your order status ...
1    \User : Agent please.\nBot: Waiting time is lo...
Name: col, dtype: object

首先，您可以将消息的拆分保存到一个var中，并使用正则表达式来完成这项工作。
所以使用纯Python，你可以尝试这样做：

messages = [
    'hello, bot',
    'How are you going?',
    'Are you here?',
    'i want know about my order status'
]
ignored = 'hello|how are you?|are you here?'
output = []

for msg in messages:
    if not re.match(ignored, msg.lower()):
        output.append(msg)

print(output)

输出应为：

['i want know about my order status']