Is it possible to map _all as a multifield?

Hello,

Is it possible to map _all as a multifield so that token stream which goes
into it would go into another field with different analyzer?

In more general case, is it possible to have a multifield which indexes its
data into another multifield so I could have dozens of fields channeling
its data into a multifield which then channels it into two multifields of
its own

for example I want to specify one multifield mapping for a field which
would channel data into stemmed and shingled multifields

I know I can do it by specifying multiple fields in the original multifield
but with several hundreds fields in my mappings I am already having hard
time managing it.

Any other approaches? I am not very eager to do templates as I would prefer
explicit and non-dynamic mapping but I may try it if there is no other
alternatives

--
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 should leave _all as it is (it is special to the include_in_all flag
in the mapping) and use your favorite fields as aggregate fields, with
using the powerful combination of "multi_field" and "index_name".

Example is bibliographic data. For my bibliographic mapping, see

In this mapping I use many Dublin Core elements as "multichannel"
aggregation fields, e.g. you can search for a title in _all, also in
"dc:creator" (all creators) but also on "xbib:namePersonal" for persons
or on "xbib:nameCorporate" for corporate names - while in the JSON
source, there was only one text string passed for indexing.

You can also see how I create non-analyzed tokens by multi_field for
faceting.

I hope this helps?

Jörg

Am 07.03.13 21:23, schrieb AlexR:

Hello,

Is it possible to map _all as a multifield so that token stream which
goes into it would go into another field with different analyzer?

In more general case, is it possible to have a multifield which
indexes its data into another multifield so I could have dozens of
fields channeling its data into a multifield which then channels it
into two multifields of its own

for example I want to specify one multifield mapping for a field which
would channel data into stemmed and shingled multifields

I know I can do it by specifying multiple fields in the original
multifield but with several hundreds fields in my mappings I am
already having hard time managing it.

Any other approaches? I am not very eager to do templates as I would
prefer explicit and non-dynamic mapping but I may try it if there is
no other alternatives

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.

Thanks Jörg,

Very helpful. I really liked how you use name spaces and your naming
conventions! I think I'll learn from it :slight_smile: will spend some time on it!

I do understand basic multifield mapping concept and use it extensively may
be I missed something but
I was just wondering if there is any way to make mapping more concise. If I
take time to declare a hundred of fields as multifield so I can combine
them together, I would love to not to have to do it 3 times once for
stemmed another for shingled and another for something else

Alex

On Thursday, March 7, 2013 4:03:04 PM UTC-5, Jörg Prante wrote:

You should leave _all as it is (it is special to the include_in_all flag
in the mapping) and use your favorite fields as aggregate fields, with
using the powerful combination of "multi_field" and "index_name".

Example is bibliographic data. For my bibliographic mapping, see
https://gist.github.com/jprante/5111676

In this mapping I use many Dublin Core elements as "multichannel"
aggregation fields, e.g. you can search for a title in _all, also in
"dc:creator" (all creators) but also on "xbib:namePersonal" for persons
or on "xbib:nameCorporate" for corporate names - while in the JSON
source, there was only one text string passed for indexing.

You can also see how I create non-analyzed tokens by multi_field for
faceting.

I hope this helps?

Jörg

Am 07.03.13 21:23, schrieb AlexR:

Hello,

Is it possible to map _all as a multifield so that token stream which
goes into it would go into another field with different analyzer?

In more general case, is it possible to have a multifield which
indexes its data into another multifield so I could have dozens of
fields channeling its data into a multifield which then channels it
into two multifields of its own

for example I want to specify one multifield mapping for a field which
would channel data into stemmed and shingled multifields

I know I can do it by specifying multiple fields in the original
multifield but with several hundreds fields in my mappings I am
already having hard time managing it.

Any other approaches? I am not very eager to do templates as I would
prefer explicit and non-dynamic mapping but I may try it if there is
no other alternatives

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.

Note, my namespace prefix use is similar to JSON-LD, but I need to
experiment with it, it should be something like in

You are right that writing mapping should be possible in a more concise
way. But then, you write a good mapping only once. So I think you can
live with it. Since a recent ES version you can also use JSON comments
/*... */ in your mapping JSON now for explaining non-obvious things
(which I couldn't back in mid 2012).

Jörg

Am 07.03.13 23:00, schrieb AlexR:

Very helpful. I really liked how you use name spaces and your naming
conventions! I think I'll learn from it :slight_smile: will spend some time on it!

I do understand basic multifield mapping concept and use it
extensively may be I missed something but
I was just wondering if there is any way to make mapping more concise.
If I take time to declare a hundred of fields as multifield so I can
combine them together, I would love to not to have to do it 3 times
once for stemmed another for shingled and another for something else

--
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.