Why if I'm placing a lookbehind constraint on the capturing group, does it ensure compliance but also capture what is prior to the given constraint?

Regex pattern in Python not capturing the correct substring, giving unexpected output

import re

#example
input_text = "It is close to that place, the NY hospital was the place where I was born, the truth is that's all I know, and it happened in November of the year 2000."

#regex pattern
# partner_match = re.search(r"(?:(?:[^.,;\n]+)(?<=\s)|^)\s*(.+?)\s*(?:was|would be|is)\s*the\s*(?:place|side)", input_text, flags = re.IGNORECASE) # here I tried using the negation operator [^...] but it doesn't work
partner_match = re.search(r"(?:(?:\.|,|;|\n)(?<=\s)|^)\s*(.+?)\s*(?:was|would be|is)\s*the\s*(?:place|side)", input_text, flags=re.IGNORECASE)

#here print captured string
if partner_match: print(partner_match.group(1))

Why instead of giving me only this output :

the NY hospital

It incorrectly gives me all this string:

It is close to that place, the NY hospital

What should I fix in my regex capture restrictions?

As explained in the comments, the regex engine will initially attempt a match at the beginning of the string. After failing to match (?:\.|,|;|\n)(?<=\s) at that location it attempts to match ^ and succeeds. It therefore will make no further attempts to match (?:\.|,|;|\n)(?<=\s), hence the undesired result.

As an aside, note that (?:\.|,|;|\n)(?<=\s) is a faulty construct as it can only match a newline (\n), never a period, comma or semicolon. It reads, "match a period, comma, semicolon or newline, provided the character following is preceded by a whitespace character", but of course the character that precedes the character that follows X is X itself. Here the newline is the only character of the four that is a whitespace character.

Another problem is that by using \s* rather than \s+ the string "wasthesize" (for example) is matched by the regular expression, \s*(?:was|would be|is)\s*the\s*(?:place|side), which I assume is an undesirable behaviour.

Note that (?:\.|,|;|\n) can be expressed more compactly as a character class: [.,;\n].

I assume the problem is to match a substring that begins after a space following a period, comma, semicolon or whitespace, and continues until it is followed by " was the place", " was the side", " would be the place", " would be the side", " is the place" or " is the side".

In the event that the string contains more than one space that is preceded by a period, comma, semicolon or whitespace, it must be decided which identifies the beginning of the matched string. I have assumed it is the last such pairing that meets all the requirements.

You therefore may attempt to match the following regular expression.

[.,;\n] (.+) +(?:was|would be|is) +the +(?:place|side)

where the desired result would be contained in capture group 1.

Demo

One may instead use a positive lookbehind and a positive lookahead to simply match (but not capture) the desired string.

(?<=[.,;\n] ).+(?= +(?:was|would be|is) +the +(?:place|side))

Demo