The current approach which specifies stopwords as
lemma: [POS]
presents two issues:
- There are some terms which POS taggers will fail over. For example, Spacy labels "AI" (artificial intelligence) as PROPN
- If I create software to be used by people without linguistic knowledge, I cannot expect them to know about POS.
As a work-around, it is necessary to specify all POS tags, which is rather inelegant.
The current approach which specifies stopwords as
lemma: [POS]
presents two issues:
As a work-around, it is necessary to specify all POS tags, which is rather inelegant.