partial matching of an entire word using elasticsearch - matching the end or middle part - elasticsearch

Question

i've read through so much of the documentation, but now getting a bit confused about how to match part of a word in a search. i understand there are many techniques, but most talk about matching the first part of a word. such as 'quick' can match 'quick brown fox'.
well, what if i have the word 'endgame' i'm looking for but the input query is 'game'? i've tried using the standard, keyword, whitespace, etc, tokenizers but i'm not getting it.
i'm sure i'm missing something simple.
update
i was able to implement this with the help of john. here's the implementation using Nest...
var ngramTokenFilter = new NgramTokenFilter
{
MinGram = 2,
MaxGram = 3
};
var nGramTokenizer = new NGramTokenizer
{
MinGram = 2, MaxGram = 3, TokenChars = new List<string>{"letter", "digit"}
};
var nGramAnalyzer = new CustomAnalyzer
{
Tokenizer = "nGramTokenizer",
Filter = new[] { "ngram", "standard", "lowercase" }
};
client.CreateIndex("myindex", i =>
{
i
.Analysis(a => a.Analyzers(an => an
.Add("ngramAnalyer", nGramAnalyzer)
)
.Tokenizers(tkn => tkn
.Add("nGramTokenizer", nGramTokenizer)
)
.TokenFilters(x => x
.Add("ngram", ngramTokenFilter)
)
)
...
and my poco, i'm actually creating a multifield, one not analyzed, and one with my ngram tokenizer analyzer:
pm.Properties(props => props
.MultiField(mf => mf
.Name("myfield")
.Fields(f => f
.String(s => s.Name("myfield").Analyzer("ngramAnalyer"))
.String(s => s.Name("raw").Index(FieldIndexOption.not_analyzed))
)
)
);

Answer 1

I would try an ngram tokenizer: http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/analysis-ngram-tokenizer.html
The example below is fairly extreme (it creates 2 and 3 letter tokens) but should give you an idea of how it works:
curl -XPUT 'localhost:9200/test' -d '
{
"settings" : {
"analysis" : {
"analyzer" : {
"my_ngram_analyzer" : {
"tokenizer" : "my_ngram_tokenizer"
}
},
"tokenizer" : {
"my_ngram_tokenizer" : {
"type" : "nGram",
"min_gram" : "2",
"max_gram" : "3",
"token_chars": [ "letter", "digit" ]
}
}
}
}
}'
curl 'localhost:9200/test/_analyze?pretty=1&analyzer=my_ngram_analyzer' -d 'FC Schalke 04'
Result: FC, Sc, Sch, ch, cha, ha, hal, al, alk, lk, lke, ke, 04
This will allow you to break up your tokens into smaller tokens of a configurable size and search on them. You'll want to play with min_gram and max_gram for your use case.
This can have some memory impact but tends to be a lot faster than a wildcard search that has trailing or leading wildcards (or both). http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/query-dsl-wildcard-query.html