Complex search terms


(Nora Olsen) #1

Hi,

I'm very new to Elastic Search or search indexes in general.

We are using haystack and have the search index mapped to our model with

  • text (title + description)
  • country name
  • city name
  • array of tags

I can see that in the index, _source is properly populated.

With this, I can perform a search on the text easily that is provided for
by ES.

But what techniques can I use this perform searches such as this
"concerts in UK"
"hiking in brazil"

or even with tags
"museums in new york" , where museums=tag and new york=city

How do people power their search bar to perform more complex searches?

--
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.


(Alexander Reelsen) #2

Hey,

as there is no named entity recognition built into elasticsearch (which
detects and extracts museum as a location for tagging and new york as a
city), you have to options:

  • Execute this named entity extraction in your application (quite some work)
  • Use the all field for this kind of queries. See all field contains the
    concatenated content of all your other fields, and will therefore match, if
    you search for 'museum in new york'. For more information about the all
    field, see http://www.elasticsearch.org/guide/reference/mapping/all-field/

If both doesnt match your requirements, feel free add some more infos and
I'll try to think harder.

--Alex

On Thu, Sep 19, 2013 at 5:51 AM, Nora Olsen nora.olsen77@gmail.com wrote:

Hi,

I'm very new to Elastic Search or search indexes in general.

We are using haystack and have the search index mapped to our model with

  • text (title + description)
  • country name
  • city name
  • array of tags

I can see that in the index, _source is properly populated.

With this, I can perform a search on the text easily that is provided for
by ES.

But what techniques can I use this perform searches such as this
"concerts in UK"
"hiking in brazil"

or even with tags
"museums in new york" , where museums=tag and new york=city

How do people power their search bar to perform more complex searches?

--
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.


(Nora Olsen) #3

Thanks. I did think about using NLP as a pre processor.

Would I be using boost to improve the score if the "museum" is a match in
the list of tags and/or "New York" in the city field, instead of some text
that has both words in the article description. I'm trying to improve the
scores of articles that have a direct match for either the tags or
location.

On Thursday, September 19, 2013 3:22:57 PM UTC+8, Alexander Reelsen wrote:

Hey,

as there is no named entity recognition built into elasticsearch (which
detects and extracts museum as a location for tagging and new york as a
city), you have to options:

  • Execute this named entity extraction in your application (quite some
    work)
  • Use the all field for this kind of queries. See all field contains the
    concatenated content of all your other fields, and will therefore match, if
    you search for 'museum in new york'. For more information about the all
    field, see http://www.elasticsearch.org/guide/reference/mapping/all-field/

If both doesnt match your requirements, feel free add some more infos and
I'll try to think harder.

--Alex

On Thu, Sep 19, 2013 at 5:51 AM, Nora Olsen <nora.o...@gmail.com<javascript:>

wrote:

Hi,

I'm very new to Elastic Search or search indexes in general.

We are using haystack and have the search index mapped to our model with

  • text (title + description)
  • country name
  • city name
  • array of tags

I can see that in the index, _source is properly populated.

With this, I can perform a search on the text easily that is provided for
by ES.

But what techniques can I use this perform searches such as this
"concerts in UK"
"hiking in brazil"

or even with tags
"museums in new york" , where museums=tag and new york=city

How do people power their search bar to perform more complex searches?

--
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.


(system) #4