// Data example
{
	"labels" : [ // NOTE: order matters
		"en:fist", // First entry of the language is the label, "en:" can be used to indicate no label in case an aliase exists
		"en:hand", // others are aliases
		"fr:poing", // label
		"fr:poignet" // aliases
	},
}

// Mapping example
labels: {
	"properties": {
		"near_match": "case insentive, accent insensitive"
		"exact_match": "case sensitive, accent sensitive" (maybe not needed)
		"prefix": "case insensitive, accenten insensitive"
		"all": { // In every subfield analyzer of this all field use a pattern char filter to remove the language prefix
			// TODO: verify that we can do it with regex char filter otherwize we may need a custom analyzer in the extra plugin
			"properties": {
				"near_match": "case insentive, accent insensitive"
				"exact_match": "case sensitive, accent sensitive" (maybe not needed)
				"prefix": "case insensitive, accenten insensitive"
			}
		}
	}
},


I'm ht, searching for poignet

// labels.all.prefix should be the union of everything else
FILTER { labels.all.prefix:poignet }
// Other fields are only needed for ranking
SHOULD DISMAX{
	// DISMAX allows to keep only the best score and not not additive scoring of all the matches
	// Might be easier when it comes to combine this score to query independent factors
	labels.all.exact_match:poignet^10
	labels.all.near_match:poignet^5

	labels.exact_match:ht:poignet^100
	labels.near_match:ht:poignet^50
	labels.prefix:ht:poignet^25

	labels.exact_match:fr:poignet^100
	labels.near_match:fr:poignet^50
	labels.prefix:fr:poignet^25

	# Should only be needed if a fallback to en
	# is interesting for ranking
	labels.exact_match:en:poignet^100
	labels.near_match:en:poignet^50
	labels.prefix:en:poignet^25
}

+ rescoring with query independant factors (sitelink...)