Add a New `WordNoMatch` Descriptor to Evidently

[elenasamuylova]

Add a New WordNoMatch Descriptor to Evidently

About Hacktoberfest contributions: https://github.com/evidentlyai/evidently/wiki/Hacktoberfest-2024

Description:

Evidently already has an ExcludesWords() descriptor that checks if the text does not contain any specified words from a shared list.

However, sometimes you might need to check that the text does not contain words specific to each row instead of a shared list.

Example:

Question	Response	Forbidden Words
"Can I cancel my subscription at any time?"	"You are allowed to cancel at any time, and we guarantee that you will receive a refund."	["guarantee", "allowed", "refund"]

What to Implement:

The new WordNoMatch() descriptor should:

1. Accept a with_column parameter: This column contains a list of forbidden words specific to each row.
2. Accept a lemmatize parameter (default True): When True, this will consider inflected or variant forms of words. Works the same as in the ExcludesWords() descriptor.
3. Return True/False:
   • Return True if the words from the list are not present in the row's text.
   • Return False if forbidden words are present.

References:

• Check the ExcludesWords() descriptor as reference.
• For a two-column descriptor implementation, check the SemanticSimilarity descriptor and the CustomPairColumnEval template.