Annotating whole sentences (without using regex)

[goonhoon]

Hi!

I am trying to write functions that will label spans (whole sentences) containing certain words or phrases.

I would normally achieve this with regex but I wonder if this is possible to do with Skweak (eg with Heuristics) in a more elegant way.

Many thanks!

[ken-dwyer]

If you are assigning a label to the whole sentence, isn't that the same as sentence classification? See for example: https://github.com/NorskRegnesentral/skweak/tree/main/examples/sentiment

[plison]

Yes, in your case I would suggest to view it as a text/span classification problem. You can simply define the span and their labels in the usual way with Skweak labelling functions, and then use NaiveBayes to aggregate the labels over your sentences.