Think about possibly enforcing uniqueItems in arrays

[geoffturk]

The idea is to avoid having this (two "bananas"):
["apple", "banana", "carrot", "pineapple", "banana"]

But what about spelling errors?
["apple", "banana", "carrot", "pineapple", "bananna"]

Or plurals?
["apple", "banana", "carrot", "pineapple", "bananas"]

Or punctuation?
["apple", "green apple", "green_apple", "green-apple"]

This brings up a few questions:

1. Should we even allow dashes and underscores, or should it only be alphanumeric characters and spaces?
2. Where do we draw the line on fuzzy matching thresholds to cater for spelling errors/variances?
3. Do we keep it simple and only filter exact matches?
4. If we decide to allow exact duplicates, how does this affect Elasticsearch's ranking (e.g., does ["apple", "apple", "banana"] come up before ["apple", "banana"] when searching for "apple"?