Facet query with custom counts in facet navigation

Jaromir_Muller · December 1, 2012, 10:41pm

Hi everyone,

I need to use facet navigation for filtering search result. By adding each
term I can build query filter like* (Attribute1=term1 OR Attribute1=term2)
AND (Attribute2=term4)*. So when I have such query, then in facet
navigation I'd like to have all available terms with counts showing
increase with new possible increase in final result set, in short user
would see how many new documents will be added to search result set.

Example:

Applied filter (Attribute1=term1 OR Attribute1=term2) AND
(Attribute2=term4)

Facet navigation

Attribute 1
term1 (applied) - count here is 0, because I cannot add more documents
to final result
- term2 (applied) - count here is 0, because I cannot add more
  documents to final result
- term3 - count would be the number of matching documents with filter
  Attribute2 = term4 AND Attribute=term3
Attribute 2
- term4 (applied) - count here is 0, because I cannot add more
  documents to final result
- term5 - count would be the number of matching documents with filter
  (Attribute1=term1 OR Attribute1=term2) AND Attribute2=term5

Can anyone advice how would search request to ES look like?

Thanks, Jaro.

--

Jaromir_Muller · December 2, 2012, 10:23am

Here an example how to replicate it and what I'd like to receive.

Thanks a lot. Jaro

On Sunday, December 2, 2012 12:41:06 AM UTC+2, Jaromír Müller wrote:

Hi everyone,

I need to use facet navigation for filtering search result. By adding each
term I can build query filter like* (Attribute1=term1 OR
Attribute1=term2) AND (Attribute2=term4)*. So when I have such query,
then in facet navigation I'd like to have all available terms with counts
showing increase with new possible increase in final result set, in short
user would see how many new documents will be added to search result set.

Example:

Applied filter (Attribute1=term1 OR Attribute1=term2) AND
(Attribute2=term4)

Facet navigation

Attribute 1

term1 (applied) - count here is 0, because I cannot add more
documents to final result

term2 (applied) - count here is 0, because I cannot add more
documents to final result

term3 - count would be the number of matching documents with
filter Attribute2 = term4 AND Attribute=term3

Attribute 2

term4 (applied) - count here is 0, because I cannot add more
documents to final result

term5 - count would be the number of matching documents with
filter (Attribute1=term1 OR Attribute1=term2) AND Attribute2=term5

Can anyone advice how would search request to ES look like?

Thanks, Jaro.

--

Jaromir_Muller · December 2, 2012, 7:53pm

In the end I found the solution which is based on facet_filter

curl -XGET 'http://localhost:9200/test_index/test_type/_search?pretty=true' -d
'{

"query":{
    "filtered":{
        "query" : {
            "query_string" : {
                "query" : "Lorem"
            }
        },
        "filter": {
            "and":[
                {
                    "or":[{
                        "term":{"size":"L"}
                    }]
                },{
                    "or":[{
                        "term":{"yesno":"yes"}
                    }]
                }
            ]
        }
    }
},
"facets":{
    "_facet_size":{
        "terms":{ 
            "field":"size",
            "order":"reverse_count",
            "all_terms":true             
        },                 
        "facet_filter": {
            "and":[
                {
                    "query" : {
                    "query_string" : {
                        "query" : "Lorem"
                    }
                }
                },
                {
                    "or":[{
                        "term": {"yesno": "yes"}
                    }]
                }
            ]
        },
        "global": true
    },
    "_facet_yesno":{
        "terms":{ 
            "field":"yesno",
            "order":"reverse_count",
            "all_terms":true             
        },                 
        "facet_filter": {
            "and":[
                {
                    "query" : {
                    "query_string" : {
                        "query" : "Lorem"
                    }
                }
                },
                {
                    "or":[{
                        "term": {"size": "L"}
                    }]
                }
            ]
        },
        "global": true
    }
}

}'

From the documentation page
(Elasticsearch Platform — Find real-time answers at scale | Elastic) wasnt
quite obvious what query I need to sent to ES.

Could anyone advice if my solution is correct or is there any possible
improvement?

Thanks, Jaro.

On Sunday, December 2, 2012 12:41:06 AM UTC+2, Jaromír Müller wrote:

Hi everyone,

I need to use facet navigation for filtering search result. By adding each
term I can build query filter like* (Attribute1=term1 OR
Attribute1=term2) AND (Attribute2=term4)*. So when I have such query,
then in facet navigation I'd like to have all available terms with counts
showing increase with new possible increase in final result set, in short
user would see how many new documents will be added to search result set.

Example:

Applied filter (Attribute1=term1 OR Attribute1=term2) AND
(Attribute2=term4)

Facet navigation

Attribute 1

term1 (applied) - count here is 0, because I cannot add more
documents to final result

term2 (applied) - count here is 0, because I cannot add more
documents to final result

term3 - count would be the number of matching documents with
filter Attribute2 = term4 AND Attribute=term3

Attribute 2

term4 (applied) - count here is 0, because I cannot add more
documents to final result

term5 - count would be the number of matching documents with
filter (Attribute1=term1 OR Attribute1=term2) AND Attribute2=term5

Can anyone advice how would search request to ES look like?

Thanks, Jaro.

--

Igor_Motov · December 3, 2012, 5:35pm

I cannot think of any other approach besides using filter facets. Facets
are typically used to narrow searches instead of expanding them. In your
case you are trying to expand your search and that means you have to run
all these extra searches for each "facet". You could also use multisearch
to do the same thing, but your solution will probably use caches better.

On Sunday, December 2, 2012 2:53:06 PM UTC-5, Jaromír Müller wrote:

In the end I found the solution which is based on facet_filter

curl -XGET '
http://localhost:9200/test_index/test_type/_search?pretty=true' -d '{
"query":{
    "filtered":{
        "query" : {
            "query_string" : {
                "query" : "Lorem"
            }
        },
        "filter": {
            "and":[
                {
                    "or":[{
                        "term":{"size":"L"}
                    }]
                },{
                    "or":[{
                        "term":{"yesno":"yes"}
                    }]
                }
            ]
        }
    }
},
"facets":{
    "_facet_size":{
        "terms":{ 
            "field":"size",
            "order":"reverse_count",
            "all_terms":true             
        },                 
        "facet_filter": {
            "and":[
                {
                    "query" : {
                    "query_string" : {
                        "query" : "Lorem"
                    }
                }
                },
                {
                    "or":[{
                        "term": {"yesno": "yes"}
                    }]
                }
            ]
        },
        "global": true
    },
    "_facet_yesno":{
        "terms":{ 
            "field":"yesno",
            "order":"reverse_count",
            "all_terms":true             
        },                 
        "facet_filter": {
            "and":[
                {
                    "query" : {
                    "query_string" : {
                        "query" : "Lorem"
                    }
                }
                },
                {
                    "or":[{
                        "term": {"size": "L"}
                    }]
                }
            ]
        },
        "global": true
    }
}
}'

From the documentation page (
Elasticsearch Platform — Find real-time answers at scale | Elastic) wasnt
quite obvious what query I need to sent to ES.

Could anyone advice if my solution is correct or is there any possible
improvement?

Thanks, Jaro.

On Sunday, December 2, 2012 12:41:06 AM UTC+2, Jaromír Müller wrote:

Hi everyone,

I need to use facet navigation for filtering search result. By adding
each term I can build query filter like* (Attribute1=term1 OR
Attribute1=term2) AND (Attribute2=term4)*. So when I have such query,
then in facet navigation I'd like to have all available terms with counts
showing increase with new possible increase in final result set, in short
user would see how many new documents will be added to search result set.

Example:

Applied filter (Attribute1=term1 OR Attribute1=term2) AND
(Attribute2=term4)

Facet navigation

Attribute 1

term1 (applied) - count here is 0, because I cannot add more
documents to final result

term2 (applied) - count here is 0, because I cannot add more
documents to final result

term3 - count would be the number of matching documents with
filter Attribute2 = term4 AND Attribute=term3

Attribute 2

term4 (applied) - count here is 0, because I cannot add more
documents to final result

term5 - count would be the number of matching documents with
filter (*Attribute1=term1 OR Attribute1=term2) AND Attribute2=term5

Can anyone advice how would search request to ES look like?

Thanks, Jaro.

--

Topic		Replies	Views
Faceted search counts for multi-valued terms Elasticsearch	3	480	July 6, 2017
Adaptive faceted searching Elasticsearch	3	537	July 6, 2017
Facet_filter and document query return different values.? Elasticsearch	2	311	July 6, 2017
Term aggregation/facet NOT INCLUDING the faceted field Elasticsearch	5	707	July 6, 2017
ElasticSearch: restrict results but not facet counts Elasticsearch	3	418	July 6, 2017

Facet query with custom counts in facet navigation

Related topics