A regular expression is a compact pattern for finding character sequences in text. It can recognize a greeting, identify a sentence boundary, or extract a value from an otherwise unstructured string. Think of a combination lock: a particular sequence makes it respond. A regular expression gives a program more flexible rules for which sequences count as a match. section 1.4.1 section 1.4.2
Patterns are useful when the form of the input is fairly predictable. In a chatbot, a match can trigger a scripted reply, while a captured part of the match can supply a name or other detail. Unlike a list containing every spelling of “Hi,” one pattern can accommodate repeated letters and several greeting forms. It still cannot anticipate every word, typo, or punctuation choice a person might use. section 1.4.2 section 1.5.2
Recognize a greeting
In Python, `import re; r = r'(hi|hello|hey)[ ,:.!]*([a-z]*)'; match = re.match(r, 'Hello Rosa', flags=re.IGNORECASE)` produces a match. The alternatives select a greeting, the punctuation group permits a separator, and the final group captures following letters. A chatbot can inspect captured groups before choosing its reply; a match alone does not establish what the speaker meant. section 1.5.2
Match the pattern to the job:
- For sentence segmentation, a pattern can split on punctuation followed by whitespace. Check the resulting sentences: `re.split` can consume the terminating punctuation, while periods inside technical text and quotations make boundaries difficult. section 11.2.1 section 11.2.2
- For information extraction, capture decimal latitude–longitude pairs in map URLs or prose. Predictable number formats are easier targets than names, places, and organizations without standard formats. section 11.4.1
Listing 11.2 A regular expression for GPS coordinates
>>> import re
>>> lat = r'([-]?[0-9]?[0-9][.][0-9]{2,10})'
>>> lon = r'([-]?1?[0-9]?[0-9][.][0-9]{2,10})'
>>> sep = r'[,/ ]{1,3}'
>>> re_gps = re.compile(lat + sep + lon)
>>> re_gps.findall('http://...maps/@34.0551066,-118.2496763...')
[(34.0551066, -118.2496763)]
>>> re_gps.findall("https://www.openstreetmap.org/#map=10/5.9666/116.0566")
[('5.9666', '116.0566')]
>>> re_gps.findall("Zig Zag Cafe is at 45.344, -121.9431 on my GPS.")
[('45.3440', '-121.9431')]
The Python listing combines latitude, separator, and longitude patterns, then uses `findall` to extract coordinate pairs from map links and a sentence. section 11.4.1
A pattern gives a definite match or nonmatch under its rules, not certainty about meaning. Adding rules for more language varieties takes manual work. Approximate matching can tolerate some errors; statistical NLP offers another approach when hardcoded patterns become brittle. Regular expressions express finite-state-style rules, although features in practical Python and POSIX implementations go beyond true regular grammars and need not share their halting guarantee. section 1.9 section 1.4.2 section 11.2.2
