Understanding intents and entities

Building Chatbots in Python

Alan Nichol

Co-founder and CTO, Rasa

An example

Intents

A restaurant_search can be expressed many different ways:

I'm hungry
Show me good pizza spots
I want to take my boyfriend out for sushi
- Can also be request_booking

Entities

NER = Named Entity Recognition

Regular expressions to recognize intents

Simpler than machine learning approaches
Highly computationally efficient
Drawback:
- Debugging regular expressions can become difficult

Using regular expressions

'|' is equivalent to OR

re.search(r"(hello|hey|hi)", "hey there!") is not None

True

re.search(r"(hello|hey|hi)", "which one?") is not None

True

Using regular expressions

\b matches the beginning or end of a word

re.search(r"\b(hello|hey|hi)\b", "hey there!") is not None

True

re.search(r"\b(hello|hey|hi)\b", "which one?") is not None

False

Using regex for entity recognition

pattern = re.compile('[A-Z]{1}[a-z]*')

message = """
Mary is a friend of mine,
she studied at Oxford and
now works at Google"""

pattern.findall(message)

['Mary', 'Oxford', 'Google']

Let's practice!

Building Chatbots in Python