Prevent abbreviations from stemming

[nataliashitova]

Summary

Short all-caps abbreviations should not be stemmed using general rules. This unwanted behavior leads to stemming, e.g., CTA to ctum in English.

Explanation

In consultation with Team Lingo we've decided that all all-caps words 2, 3, and 4 characters long should be prevented from stemming. The result of relevantWords researches should then include abbreviations in a shape of WordCombinations, where word is an all-caps abbreviation, stem is a lower-case abbreviation, occurrences is a number of occurrences.