# Relevance with short combinations of letters
**URL:** https://discuss.elastic.co/t/relevance-with-short-combinations-of-letters/127813
**Category:** Elasticsearch
**Created:** [April 12, 2018, 12:25pm UTC](https://discuss.elastic.co/t/relevance-with-short-combinations-of-letters/127813 "2018-04-12T12:25:11Z")
**Posts on this page:** 2
**Page:** 1
### Author:  [@Snail\_Banshee](https://discuss.elastic.co/u/Snail_Banshee)
#### Post date: [April 12, 2018, 12:25pm UTC](https://discuss.elastic.co/t/relevance-with-short-combinations-of-letters/127813/1 "2018-04-12T12:25:12Z")
Hello, everyone!🙂
I`m trying to make relevant search for cars (inc. brands models and part names),
for example search request is: "air filter audi a3".
My settings (sorry for php array):
'settings' =\> [
'analysis' =\> [
'filter'=\>[
'ru\_stop'=\>[
"type"=\>"stop",
"stopwords"=\>"_russian_"
],
'ru\_stemmer'=\>[
"type"=\>"stemmer",
"language"=\>"russian"
],
'english\_stop'=\>[
"type"=\>"stop",
"stopwords"=\>"_english_"
],
'english\_stemmer'=\>[
"type"=\>"stemmer",
"language"=\>"english"
],
'synonym'=\>[
"type"=\>"synonym",
"synonyms\_path"=\>$this-\>iCustom-\>iDRoot."/files/es\_synonyms.txt"
],
'snowball'=\>[
"type"=\>"snowball",
"language"=\>"Russian"
],
"word\_delimiter"=\>[
"type"=\>"word\_delimiter",
"split\_on\_numerics"=\>false,
"split\_on\_case\_change"=\>false,
"generate\_word\_parts"=\>true,
"generate\_number\_parts"=\>false,
"catenate\_words"=\>true,
'preserve\_original'=\>true
],
'ascii\_folding'=\>[
"type"=\>"asciifolding",
"preserve\_original"=\>true
],
'gram\_filter'=\>[
'type'=\>'nGram',
'min\_gram'=\>3,
'max\_gram'=\>15,
'token\_chars'=\>[
'letter',
'digit'
]
],
'unique\_stem'=\>[
'type'=\>'unique',
'only\_on\_same\_position'=\>true
],
'length\_filter'=\>[
'type'=\>'length',
'min'=\>2,
'max'=\>20
],
],
'tokenizer'=\>[
'gram\_tokenizer'=\>[
'type'=\>'nGram',
'min\_gram'=\>3,
'max\_gram'=\>15,
'token\_chars'=\>[
'letter',
'digit'
]
]
],
'analyzer'=\>[
'default'=\>[
'type'=\>'custom',
"char\_filter"=\>[
"html\_strip"
],
"tokenizer"=\>'standard',
"filter"=\>[
'trim',
"lowercase",
"word\_delimiter",
"snowball",
"ascii\_folding",
//"keyword\_repeat",
//"unique\_stem",
"ru\_stop",
"ru\_stemmer",
"english\_stop",
"english\_stemmer",
"synonym",
'length\_filter',
//"gram\_filter",
]
]
]
]
],
'mappings'=\>[
'item'=\>[
'properties'=\>[
"title"=\>[
"type"=\>"string",
"index"=\>"not\_analyzed",
'index\_options'=\>'docs',
],
"data"=\>[
"type"=\>"string",
'analyzer'=\>'default',
],
"description"=\>[
"type"=\>"string",
"analyzer"=\>"default",
"store"=\>true
],
"numbers"=\>[
"type"=\>"string",
"store"=\>true,
"index"=\>"not\_analyzed"
],
"timestamp"=\>[
"type"=\>"integer",
"index"=\>"not\_analyzed"
],
]
]
]
Some "data" in ES:
- Air filter MAPCO 65217; AUDI A4 (8E2, B6) 1.8 T 2002 2001 2000; AUDI A4 Avant (8E5, B6) 1.8 T 2002 2001;
- Air filter: Air filter MAGNETI MARELLI 153071760244; AUDI A4 (8K2, B8) 1.8 TFSI 2012 2011 2010 2009 2008 2007; AUDI A4 Avant (8K5, B8) 1.8 TFSI 2012 2011 2010 2009 2008 2007;
- Air filter FEBI BILSTEIN 48477; AUDI Q7 (4LB) 3.0 TDI quattro 2008 2007 2006; VW TOUAREG (7LA, 7L6, 7L7) 2.5 R5 TDI 2010 2009 2008 2007 2006 2005 2004 2003;
- Air filter: Air filter FRAM CA10236; AUDI Q7 (4LB) 3.0 TDI quattro 2010 2009 2008 2007 2006; AUDI A3 (8V1, 8VK) S3 quattro 2018 2017 2016 2015 2014 2013 2012; PORSCHE CAYENNE (9PA) 3.0 TDI 2010 2009;
- Inside air filter FRAM CFA8869; AUDI TT Roadster (8N9) 1.8 T 2006 2005 2004 2003 2002 2001 2000 1999; AUDI A3 (8L1) 1.6 2003 2002 2001 2000 1999 1998 1997 1996; AUDI TT (8N3) 1.8 T 2006 2005 2004 2003 2002 2001 2000 1999 1998;
etc...
So, i have response with score for data that contains Audi A3 lower than forexample Audi A4 or Audi Q7, etc.
Why is it so?
P.S.: i`m new at ES, so do not judge strictly:))
Thank2all!
---
### Author:  [@system](https://discuss.elastic.co/u/system)
#### Post date: [May 10, 2018, 12:25pm UTC](https://discuss.elastic.co/t/relevance-with-short-combinations-of-letters/127813/2 "2018-05-10T12:25:30Z")
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.