Issues with search on multi_field field

Derry_O_Sullivan · February 28, 2013, 9:09am

Hi all,

I'm having issues with searching the analyzed and not_analyzed fields in a
multi_field object (using ES 0.20.4):

I've created a gist here:

gist.github.com

https://gist.github.com/derryos/2218785ca960e3a4f30f

gistfile1.sh

# Create an index with some language support:
http://localhost:9200/items/_settings
{
items: {
settings: {
index.version.created: "200499",
index.number_of_replicas: "1",
index.analysis.analyzer.en.type: "english",
index.number_of_shards: "5",
index.analysis.analyzer.null.type: "standard",

This file has been truncated. show original

The gist workflow is:

Create an index with language analyzers configured
Create an index type with a multi_field value which is split into
analyzed and not_analyzed. (note the analyzer is based on another field
value called language)
Try and search for terms using the analyzed field (different results
returned depending on on which language is set for indexing/search)
Try and search for terms using the not_analyzed field (should get more
results due to lack of language analysis).
Recreate my mapping to try and force ES to store/index the 'raw' field
part of the multi_field

I've having trouble doing the searches with the unanalyzed fields - as per
kimchy's example here:

gist.github.com

https://gist.github.com/kimchy/1296043

gistfile1.sh

curl -XPUT localhost:9200/test -d '{
    "mappings" : {
        "type1" : {
           "properties" : {
               "name" : {
                   "type" : "multi_field",
                   "fields" : {
                       "name" : {"type" : "string", "index" : "analyzed"},
                       "untouched" : {"type" : "string", "index" : "not_analyzed"}
                   }

This file has been truncated. show original

This all works but if i change the last line to be:
curl -XGET localhost:9200/test/_search?q=name.untouched:me
rather than:
curl -XGET localhost:9200/test/_search?q=name.untouched:*

I get 0 results. What i want is the ability to search over the untouched
fields also with terms.

Any help greatly appreciated.

Derry

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

Derry_O_Sullivan · February 28, 2013, 9:16am

Sorry, think i may have answered my own question (a little)!

If i search for:
curl -XGET localhost:9200/test/_search?q=name.untouched:"test me"

I get the result back (i.e. the whole not_analyzed) stores the entire text
as 1 token. In my use case (if i want to expose "test" in the search and
"me" in the search, would stop/simple/whitespace be a better solution than
non analyzed?

On 28 February 2013 09:09, Derry O' Sullivan derryos@gmail.com wrote:

Hi all,

I'm having issues with searching the analyzed and not_analyzed fields in a
multi_field object (using ES 0.20.4):

I've created a gist here:
Trying to do language specific searching with multi_field in ES · GitHub

The gist workflow is:

Create an index with language analyzers configured

Create an index type with a multi_field value which is split into
analyzed and not_analyzed. (note the analyzer is based on another field
value called language)

Try and search for terms using the analyzed field (different results
returned depending on on which language is set for indexing/search)

Try and search for terms using the not_analyzed field (should get more
results due to lack of language analysis).

Recreate my mapping to try and force ES to store/index the 'raw' field
part of the multi_field

I've having trouble doing the searches with the unanalyzed fields - as per
kimchy's example here:
gist:1296043 · GitHub

This all works but if i change the last line to be:
curl -XGET localhost:9200/test/_search?q=name.untouched:me
rather than:
curl -XGET localhost:9200/test/_search?q=name.untouched:*

I get 0 results. What i want is the ability to search over the untouched
fields also with terms.

Any help greatly appreciated.

Derry

--
You received this message because you are subscribed to a topic in the
Google Groups "elasticsearch" group.
To unsubscribe from this topic, visit
https://groups.google.com/d/topic/elasticsearch/n8Z9cDpqRz4/unsubscribe?hl=en-US
.
To unsubscribe from this group and all its topics, send an email to
elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

Ivan · February 28, 2013, 6:19pm

Why not simply search on the "name" field which is analyzed?

The default (standard) analyzer will tokenize on common word boundaries,
plus lowercase and stop filters. The choice if analyzer depends on your
goals.

On Thu, Feb 28, 2013 at 1:16 AM, Derry O' Sullivan derryos@gmail.comwrote:

Sorry, think i may have answered my own question (a little)!

If i search for:
curl -XGET localhost:9200/test/_search?q=name.untouched:"test me"

I get the result back (i.e. the whole not_analyzed) stores the entire text
as 1 token. In my use case (if i want to expose "test" in the search and
"me" in the search, would stop/simple/whitespace be a better solution than
non analyzed?

On 28 February 2013 09:09, Derry O' Sullivan derryos@gmail.com wrote:

Hi all,

I'm having issues with searching the analyzed and not_analyzed fields in
a multi_field object (using ES 0.20.4):

I've created a gist here:
Trying to do language specific searching with multi_field in ES · GitHub

The gist workflow is:

Create an index with language analyzers configured

Create an index type with a multi_field value which is split into
analyzed and not_analyzed. (note the analyzer is based on another field
value called language)

Try and search for terms using the analyzed field (different results
returned depending on on which language is set for indexing/search)

Try and search for terms using the not_analyzed field (should get more
results due to lack of language analysis).

Recreate my mapping to try and force ES to store/index the 'raw' field
part of the multi_field

I've having trouble doing the searches with the unanalyzed fields - as
per kimchy's example here:
gist:1296043 · GitHub

This all works but if i change the last line to be:
curl -XGET localhost:9200/test/_search?q=name.untouched:me
rather than:
curl -XGET localhost:9200/test/_search?q=name.untouched:*

I get 0 results. What i want is the ability to search over the untouched
fields also with terms.

Any help greatly appreciated.

Derry

--
You received this message because you are subscribed to a topic in the
Google Groups "elasticsearch" group.
To unsubscribe from this topic, visit
https://groups.google.com/d/topic/elasticsearch/n8Z9cDpqRz4/unsubscribe?hl=en-US
.
To unsubscribe from this group and all its topics, send an email to
elasticsearch+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/groups/opt_out.

--
You received this message because you are subscribed to the Google Groups
"elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an
email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

Clinton_Gormley · March 1, 2013, 1:22pm

Hi Derry

I'm having issues with searching the analyzed and not_analyzed fields
in a multi_field object (using ES 0.20.4):

I've created a gist here:
Trying to do language specific searching with multi_field in ES · GitHub

You've misunderstood not_analyzed fields.

not_analyzed fields index the exact value in the field, so eg:

"The quick brown fox" ->
analyzed: ["quick","brown","fox"]
not_analyzed: ["The quick brown fox"]

So the only search that will work on the not_analyzed field is a search
for "The quick brown fox". Even "the quick brown fox" won't work,
because the case is different

clint

The gist workflow is:

Create an index with language analyzers configured

Create an index type with a multi_field value which is split into
analyzed and not_analyzed. (note the analyzer is based on another
field value called language)

Try and search for terms using the analyzed field (different
results returned depending on on which language is set for
indexing/search)

Try and search for terms using the not_analyzed field (should get
more results due to lack of language analysis).

Recreate my mapping to try and force ES to store/index the 'raw'
field part of the multi_field

I've having trouble doing the searches with the unanalyzed fields - as
per kimchy's example here:
gist:1296043 · GitHub

This all works but if i change the last line to be:
curl -XGET localhost:9200/test/_search?q=name.untouched:me

rather than:
curl -XGET localhost:9200/test/_search?q=name.untouched:*

I get 0 results. What i want is the ability to search over the
untouched fields also with terms.

Any help greatly appreciated.

Derry

--
You received this message because you are subscribed to the Google
Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

Derry_O_Sullivan · March 4, 2013, 10:34am

Hi Clint, Ivan,

Thanks for the fast response.

Clint, i thought that not_analyzed mean more 'simple' analysis so thanks
for the clarification.

My use case is that i insert content into elasticsearch based on automatic
language detection (e.g. tika, user input, or other language setting
measures). This happens automatically so i have no idea what the
index_analyzer (set on the language field) is going to be when the content
is ingested. When i go to use the content (in search), i want to use a
'best fit' analyzer or else the language analyzer a user specifies in
search. (so my real question is how do i solve someone searching in a
potentially different language to the language that some relevant content
has been inserted with)

My logic on using multi field was that i was going to have the content
analyzed in the (hopefully) correct language and then a catch all version
of the content (sounds to me like this should be standard or something
simpler?) which is language ambiguous.

Therefore if i go to search the content using a search analyzer in a
different language to the index analyzer, i could have some chance of
getting some text matching (although i'm guessing that i would need to do 2
searches - one with the 'best fit' analyzer' and one with the catch all
analyzer?

Thanks again,

Derry

On Friday, 1 March 2013 13:22:50 UTC, Clinton Gormley wrote:

Hi Derry

I'm having issues with searching the analyzed and not_analyzed fields
in a multi_field object (using ES 0.20.4):

I've created a gist here:
Trying to do language specific searching with multi_field in ES · GitHub

You've misunderstood not_analyzed fields.

not_analyzed fields index the exact value in the field, so eg:

"The quick brown fox" ->
analyzed: ["quick","brown","fox"]
not_analyzed: ["The quick brown fox"]

So the only search that will work on the not_analyzed field is a search
for "The quick brown fox". Even "the quick brown fox" won't work,
because the case is different

clint

The gist workflow is:

Create an index with language analyzers configured

Create an index type with a multi_field value which is split into
analyzed and not_analyzed. (note the analyzer is based on another
field value called language)

Try and search for terms using the analyzed field (different
results returned depending on on which language is set for
indexing/search)

Try and search for terms using the not_analyzed field (should get
more results due to lack of language analysis).

Recreate my mapping to try and force ES to store/index the 'raw'
field part of the multi_field

I've having trouble doing the searches with the unanalyzed fields - as
per kimchy's example here:
gist:1296043 · GitHub

This all works but if i change the last line to be:
curl -XGET localhost:9200/test/_search?q=name.untouched:me

rather than:
curl -XGET localhost:9200/test/_search?q=name.untouched:*

I get 0 results. What i want is the ability to search over the
untouched fields also with terms.

Any help greatly appreciated.

Derry

--
You received this message because you are subscribed to the Google
Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to elasticsearc...@googlegroups.com <javascript:>.
For more options, visit https://groups.google.com/groups/opt_out.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

Clinton_Gormley · March 4, 2013, 10:44am

My logic on using multi field was that i was going to have the content
analyzed in the (hopefully) correct language and then a catch all
version of the content (sounds to me like this should be standard or
something simpler?) which is language ambiguous.

So yes - use one field with just the "standard" analyzer.

Therefore if i go to search the content using a search analyzer in a
different language to the index analyzer, i could have some chance of
getting some text matching (although i'm guessing that i would need to
do 2 searches - one with the 'best fit' analyzer' and one with the
catch all analyzer?

You don't need two queries, just two query clauses. Or even a
multi_match query that looks at both fields, and can assign them
different boosts

clint

Thanks again,

Derry

On Friday, 1 March 2013 13:22:50 UTC, Clinton Gormley wrote:
Hi Derry
>
>
> I'm having issues with searching the analyzed and
not_analyzed fields
> in a multi_field object (using ES 0.20.4):
>
>
> I've created a gist here:
> Trying to do language specific searching with multi_field in ES · GitHub

    You've misunderstood not_analyzed fields. 
    
    not_analyzed fields index the exact value in the field, so
    eg: 
    
       "The quick brown fox" -> 
        analyzed:      ["quick","brown","fox"] 
        not_analyzed:  ["The quick brown fox"] 
    
    So the only search that will work on the not_analyzed field is
    a search 
    for "The quick brown fox".  Even "the quick brown fox" won't
    work, 
    because the case is different 
    
    clint 
      
    > 
    > 
    > 
    > The gist workflow is: 
    > 1. Create an index with language analyzers configured 
    > 2. Create an index type with a multi_field value which is
    split into 
    > analyzed and not_analyzed. (note the analyzer is based on
    another 
    > field value called language) 
    > 3. Try and search for terms using the analyzed field
    (different 
    > results returned depending on on which language is set for 
    > indexing/search) 
    > 4. Try and search for terms using the not_analyzed field
    (should get 
    > more results due to lack of language analysis). 
    > 5. Recreate my mapping to try and force ES to store/index
    the 'raw' 
    > field part of the multi_field 
    > 
    > 
    > I've having trouble doing the searches with the unanalyzed
    fields - as 
    > per kimchy's example here: 
    > https://gist.github.com/kimchy/1296043 
    > 
    > 
    > 
    > This all works but if i change the last line to be: 
    > curl -XGET localhost:9200/test/_search?q=name.untouched:me 
    > 
    > rather than: 
    > curl -XGET localhost:9200/test/_search?q=name.untouched:* 
    > 
    > 
    > 
    > I get 0 results. What i want is the ability to search over
    the 
    > untouched fields also with terms. 
    > 
    > 
    > 
    > Any help greatly appreciated. 
    > 
    > 
    > Derry 
    > 
    > -- 
    > You received this message because you are subscribed to the
    Google 
    > Groups "elasticsearch" group. 
    > To unsubscribe from this group and stop receiving emails
    from it, send 
    > an email to elasticsearc...@googlegroups.com. 
    > For more options, visit
    https://groups.google.com/groups/opt_out. 
    >   
    >

--
You received this message because you are subscribed to the Google
Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

Derry_O_Sullivan · March 4, 2013, 11:32am

Thanks for the fast response!

So (just to clarify) you are saying to index in both language driven and
standard analyzer e.g.:

and then search (using multimatch) in both language driven and the standard
analyzer with a boost on the language driven query within the multimatch?

e.g:

curl -XPUT 'http://localhost:9200/items/testLanguage/_mapping' -d'{
"testLanguage" : {
"_analyzer" : {
"path" : "language"
},
"properties" : {
"name" : {
"type" : "multi_field",
"fields" : {
"name" : {"type" : "string", "index" : "analyzed"},
"untouched" : {"type" : "string", "index" : "analyzed", "index_analyzer" : "standard"}
}
},
"language" : {
"type" : "string",
"index":"not_analyzed"
}
}
}
}'

and then in the search do something like:

{
"multi_match" : {
"query" : ,
"fields" : [ "name, "untouched" ]
}
}

?

Cheers,

Derry

On Monday, 4 March 2013 10:44:32 UTC, Clinton Gormley wrote:

My logic on using multi field was that i was going to have the content
analyzed in the (hopefully) correct language and then a catch all
version of the content (sounds to me like this should be standard or
something simpler?) which is language ambiguous.

So yes - use one field with just the "standard" analyzer.

Therefore if i go to search the content using a search analyzer in a
different language to the index analyzer, i could have some chance of
getting some text matching (although i'm guessing that i would need to
do 2 searches - one with the 'best fit' analyzer' and one with the
catch all analyzer?

You don't need two queries, just two query clauses. Or even a
multi_match query that looks at both fields, and can assign them
different boosts

clint
Thanks again,

Derry

On Friday, 1 March 2013 13:22:50 UTC, Clinton Gormley wrote:
Hi Derry
>
>
> I'm having issues with searching the analyzed and
not_analyzed fields
> in a multi_field object (using ES 0.20.4):
>
>
> I've created a gist here:
> Trying to do language specific searching with multi_field in ES · GitHub
    You've misunderstood not_analyzed fields. 
    
    not_analyzed fields index the exact value in the field, so 
    eg: 
    
       "The quick brown fox" -> 
        analyzed:      ["quick","brown","fox"] 
        not_analyzed:  ["The quick brown fox"] 
    
    So the only search that will work on the not_analyzed field is 
    a search 
    for "The quick brown fox".  Even "the quick brown fox" won't 
    work, 
    because the case is different 
    
    clint 
      
    > 
    > 
    > 
    > The gist workflow is: 
    > 1. Create an index with language analyzers configured 
    > 2. Create an index type with a multi_field value which is 
    split into 
    > analyzed and not_analyzed. (note the analyzer is based on 
    another 
    > field value called language) 
    > 3. Try and search for terms using the analyzed field 
    (different 
    > results returned depending on on which language is set for 
    > indexing/search) 
    > 4. Try and search for terms using the not_analyzed field 
    (should get 
    > more results due to lack of language analysis). 
    > 5. Recreate my mapping to try and force ES to store/index 
    the 'raw' 
    > field part of the multi_field 
    > 
    > 
    > I've having trouble doing the searches with the unanalyzed 
    fields - as 
    > per kimchy's example here: 
    > https://gist.github.com/kimchy/1296043 
    > 
    > 
    > 
    > This all works but if i change the last line to be: 
    > curl -XGET localhost:9200/test/_search?q=name.untouched:me 
    > 
    > rather than: 
    > curl -XGET localhost:9200/test/_search?q=name.untouched:* 
    > 
    > 
    > 
    > I get 0 results. What i want is the ability to search over 
    the 
    > untouched fields also with terms. 
    > 
    > 
    > 
    > Any help greatly appreciated. 
    > 
    > 
    > Derry 
    > 
    > -- 
    > You received this message because you are subscribed to the 
    Google 
    > Groups "elasticsearch" group. 
    > To unsubscribe from this group and stop receiving emails 
    from it, send 
    > an email to elasticsearc...@googlegroups.com. 
    > For more options, visit 
    https://groups.google.com/groups/opt_out. 
    >   
    >   
--
You received this message because you are subscribed to the Google
Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to elasticsearc...@googlegroups.com <javascript:>.
For more options, visit https://groups.google.com/groups/opt_out.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

Clinton_Gormley · March 4, 2013, 12:04pm

{
"multi_match" : {
"query" : ,
"fields" : [ "name, "untouched" ]
}
}

The field would be "name.untouched", but otherwise yes. Also, you can
boost individual fields like:

"field": ["name^2","name.untouched"]

clint

?

Cheers,

Derry

On Monday, 4 March 2013 10:44:32 UTC, Clinton Gormley wrote:

    > 
    > My logic on using multi field was that i was going to have
    the content 
    > analyzed in the (hopefully) correct language and then a
    catch all 
    > version of the content (sounds to me like this should be
    standard or 
    > something simpler?) which is language ambiguous. 
    
    So yes - use one field with just the "standard" analyzer. 
    
    > 
    > 
    > Therefore if i go to search the content using a search
    analyzer in a 
    > different language to the index analyzer, i could have some
    chance of 
    > getting some text matching (although i'm guessing that i
    would need to 
    > do 2 searches - one with the 'best fit' analyzer' and one
    with the 
    > catch all analyzer? 
    
    You don't need two queries, just two query clauses.  Or even
    a 
    multi_match query that looks at both fields, and can assign
    them 
    different boosts 
    
    clint 
    
    > 
    > 
    > Thanks again, 
    > 
    > 
    > Derry 
    > 
    > 
    > 
    > 
    > 
    > On Friday, 1 March 2013 13:22:50 UTC, Clinton Gormley
    wrote: 
    >         Hi Derry 
    >         > 
    >         > 
    >         > I'm having issues with searching the analyzed and 
    >         not_analyzed fields 
    >         > in a multi_field object (using ES 0.20.4): 
    >         > 
    >         > 
    >         > I've created a gist here: 
    >         >
    https://gist.github.com/derryos/2218785ca960e3a4f30f 
    >         
    >         You've misunderstood not_analyzed fields. 
    >         
    >         not_analyzed fields index the exact value in the
    field, so 
    >         eg: 
    >         
    >            "The quick brown fox" -> 
    >             analyzed:      ["quick","brown","fox"] 
    >             not_analyzed:  ["The quick brown fox"] 
    >         
    >         So the only search that will work on the
    not_analyzed field is 
    >         a search 
    >         for "The quick brown fox".  Even "the quick brown
    fox" won't 
    >         work, 
    >         because the case is different 
    >         
    >         clint 
    >           
    >         > 
    >         > 
    >         > 
    >         > The gist workflow is: 
    >         > 1. Create an index with language analyzers
    configured 
    >         > 2. Create an index type with a multi_field value
    which is 
    >         split into 
    >         > analyzed and not_analyzed. (note the analyzer is
    based on 
    >         another 
    >         > field value called language) 
    >         > 3. Try and search for terms using the analyzed
    field 
    >         (different 
    >         > results returned depending on on which language is
    set for 
    >         > indexing/search) 
    >         > 4. Try and search for terms using the not_analyzed
    field 
    >         (should get 
    >         > more results due to lack of language analysis). 
    >         > 5. Recreate my mapping to try and force ES to
    store/index 
    >         the 'raw' 
    >         > field part of the multi_field 
    >         > 
    >         > 
    >         > I've having trouble doing the searches with the
    unanalyzed 
    >         fields - as 
    >         > per kimchy's example here: 
    >         > https://gist.github.com/kimchy/1296043 
    >         > 
    >         > 
    >         > 
    >         > This all works but if i change the last line to
    be: 
    >         > curl -XGET
    localhost:9200/test/_search?q=name.untouched:me 
    >         > 
    >         > rather than: 
    >         > curl -XGET
    localhost:9200/test/_search?q=name.untouched:* 
    >         > 
    >         > 
    >         > 
    >         > I get 0 results. What i want is the ability to
    search over 
    >         the 
    >         > untouched fields also with terms. 
    >         > 
    >         > 
    >         > 
    >         > Any help greatly appreciated. 
    >         > 
    >         > 
    >         > Derry 
    >         > 
    >         > -- 
    >         > You received this message because you are
    subscribed to the 
    >         Google 
    >         > Groups "elasticsearch" group. 
    >         > To unsubscribe from this group and stop receiving
    emails 
    >         from it, send 
    >         > an email to elasticsearc...@googlegroups.com. 
    >         > For more options, visit 
    >         https://groups.google.com/groups/opt_out. 
    >         >   
    >         >   
    >         
    >         
    > 
    > -- 
    > You received this message because you are subscribed to the
    Google 
    > Groups "elasticsearch" group. 
    > To unsubscribe from this group and stop receiving emails
    from it, send 
    > an email to elasticsearc...@googlegroups.com. 
    > For more options, visit
    https://groups.google.com/groups/opt_out. 
    >   
    >

--
You received this message because you are subscribed to the Google
Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

Derry_O_Sullivan · March 4, 2013, 12:10pm

Perfect, thanks for that!

On 4 March 2013 12:04, Clinton Gormley clint@traveljury.com wrote:

{
"multi_match" : {
"query" : ,
"fields" : [ "name, "untouched" ]
}
}

The field would be "name.untouched", but otherwise yes. Also, you can
boost individual fields like:

"field": ["name^2","name.untouched"]

clint

?

Cheers,

Derry

On Monday, 4 March 2013 10:44:32 UTC, Clinton Gormley wrote:

    >
    > My logic on using multi field was that i was going to have
    the content
    > analyzed in the (hopefully) correct language and then a
    catch all
    > version of the content (sounds to me like this should be
    standard or
    > something simpler?) which is language ambiguous.

    So yes - use one field with just the "standard" analyzer.

    >
    >
    > Therefore if i go to search the content using a search
    analyzer in a
    > different language to the index analyzer, i could have some
    chance of
    > getting some text matching (although i'm guessing that i
    would need to
    > do 2 searches - one with the 'best fit' analyzer' and one
    with the
    > catch all analyzer?

    You don't need two queries, just two query clauses.  Or even
    a
    multi_match query that looks at both fields, and can assign
    them
    different boosts

    clint

    >
    >
    > Thanks again,
    >
    >
    > Derry
    >
    >
    >
    >
    >
    > On Friday, 1 March 2013 13:22:50 UTC, Clinton Gormley
    wrote:
    >         Hi Derry
    >         >
    >         >
    >         > I'm having issues with searching the analyzed and
    >         not_analyzed fields
    >         > in a multi_field object (using ES 0.20.4):
    >         >
    >         >
    >         > I've created a gist here:
    >         >
    https://gist.github.com/derryos/2218785ca960e3a4f30f
    >
    >         You've misunderstood not_analyzed fields.
    >
    >         not_analyzed fields index the exact value in the
    field, so
    >         eg:
    >
    >            "The quick brown fox" ->
    >             analyzed:      ["quick","brown","fox"]
    >             not_analyzed:  ["The quick brown fox"]
    >
    >         So the only search that will work on the
    not_analyzed field is
    >         a search
    >         for "The quick brown fox".  Even "the quick brown
    fox" won't
    >         work,
    >         because the case is different
    >
    >         clint
    >
    >         >
    >         >
    >         >
    >         > The gist workflow is:
    >         > 1. Create an index with language analyzers
    configured
    >         > 2. Create an index type with a multi_field value
    which is
    >         split into
    >         > analyzed and not_analyzed. (note the analyzer is
    based on
    >         another
    >         > field value called language)
    >         > 3. Try and search for terms using the analyzed
    field
    >         (different
    >         > results returned depending on on which language is
    set for
    >         > indexing/search)
    >         > 4. Try and search for terms using the not_analyzed
    field
    >         (should get
    >         > more results due to lack of language analysis).
    >         > 5. Recreate my mapping to try and force ES to
    store/index
    >         the 'raw'
    >         > field part of the multi_field
    >         >
    >         >
    >         > I've having trouble doing the searches with the
    unanalyzed
    >         fields - as
    >         > per kimchy's example here:
    >         > https://gist.github.com/kimchy/1296043
    >         >
    >         >
    >         >
    >         > This all works but if i change the last line to
    be:
    >         > curl -XGET
    localhost:9200/test/_search?q=name.untouched:me
    >         >
    >         > rather than:
    >         > curl -XGET
    localhost:9200/test/_search?q=name.untouched:*
    >         >
    >         >
    >         >
    >         > I get 0 results. What i want is the ability to
    search over
    >         the
    >         > untouched fields also with terms.
    >         >
    >         >
    >         >
    >         > Any help greatly appreciated.
    >         >
    >         >
    >         > Derry
    >         >
    >         > --
    >         > You received this message because you are
    subscribed to the
    >         Google
    >         > Groups "elasticsearch" group.
    >         > To unsubscribe from this group and stop receiving
    emails
    >         from it, send
    >         > an email to elasticsearc...@googlegroups.com.
    >         > For more options, visit
    >         https://groups.google.com/groups/opt_out.
    >         >
    >         >
    >
    >
    >
    > --
    > You received this message because you are subscribed to the
    Google
    > Groups "elasticsearch" group.
    > To unsubscribe from this group and stop receiving emails
    from it, send
    > an email to elasticsearc...@googlegroups.com.
    > For more options, visit
    https://groups.google.com/groups/opt_out.
    >
    >

--
You received this message because you are subscribed to the Google
Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

--
You received this message because you are subscribed to a topic in the
Google Groups "elasticsearch" group.
To unsubscribe from this topic, visit
https://groups.google.com/d/topic/elasticsearch/n8Z9cDpqRz4/unsubscribe?hl=en-US
.
To unsubscribe from this group and all its topics, send an email to
elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

Topic		Replies	Views
Inconsistent query behavior with multi_field and string value Elasticsearch	2	340	July 6, 2017
Multi-word, multi-field search with analyzers Elasticsearch	1	395	July 6, 2017
Multi_field mapping Elasticsearch	2	371	July 6, 2017
Elasticsearch 1.0 RC 1 - strange issue with multi_fields faceting Elasticsearch	4	351	July 6, 2017
Question about multi_field and edge ngram Elasticsearch	11	602	July 6, 2017

Issues with search on multi_field field

Related topics