support for common "be" contraction vs possession

[lingdoc]

Proper nouns are left alone by the expander:
Mary's happy about Bill's job > Mary's happy about Bill's job

But here Mary's is actually a contraction of Mary and is. Compare with:
She's glad > She has glad (Incorrect but closer to a verbal construction)

Any plans to support these ubiquitous contractions by checking whether the following word is an adjective/verb?

[ian-beaver]

That is a common case, however to solve it would require introducing a POS tagger to the expansion step, which is already on the slow side, so I had decided to leave it out. I try not to do more than necessary as this library is commonly used to pre-process text for machine learning data, and we work with large volumes of it. The additional minor overhead of POS tagging compounds quickly on large datasets. That being said, I suppose I could make that an optional step controlled with a flag or add it to the precise method.

[lingdoc]

You could get quite a bit of mileage from checking against a list of common English adjectives/verbs. In the example above, if the word after a contraction is in the list, the contracted form is likely to be 'is' or some variation. I would think that checking a list/dict would reduce the overhead significantly.