How to handle "e.g.", "i.e.", "viz.", etc.

[fi11222]

The standard English models (I currently use en_core_web_trf) do not seem to be able to properly identify these common abbreviations.

Currently, they are categorized as "foreign words" (pos_ = X, tag_ = FW) and the final "." is yanked off from "viz." and treated as a separate token.

Is there something I can do to avoid that?

I did not find anything about this issue either here or elsewhere on the Web

[DuyguA]

The tag X is used to denote words that don’t clearly fit into other standard part-of-speech categories, often referred to as the "other" or "unknown" category. This tag is assigned when the model cannot confidently classify the word into a more specific category like noun, verb, adverb, etc.

For i.e. tagging it as X sounds OK to me, because it is an abbreviation and doesn’t fit neatly into conventional grammatical categories. While it functions similarly to an adverb in context, the model might default to X due to its abbreviation nature and less frequent occurrence in training corpora.