Collapse and Expand Results

The Collapsing query parser and the Expand component combine to form an approach to grouping documents for field collapsing in search results.

The Collapsing query parser groups documents (collapsing the result set) according to your parameters, while the Expand component provides access to documents in the collapsed group for use in results display or other processing by a client application. Collapse & Expand can together do what the older Result Grouping (group=true) does for most use-cases but not all. Collapse and Expand are not supported when Result Grouping is enabled. Generally, you should prefer Collapse & Expand.

In order to use these features with SolrCloud, the documents must be located on the same shard. To ensure document co-location, you can define the router.name parameter as compositeId when creating the collection. For more information on this option, see the section Document Routing.

Collapsing Query Parser

The CollapsingQParser is really a post filter that provides more performant field collapsing than Solr’s standard approach when the number of distinct groups in the result set is high. This parser collapses the result set to a single document per group before it forwards the result set to the rest of the search components. So all downstream components (faceting, highlighting, etc.) will work with the collapsed result set.

The CollapsingQParserPlugin fully supports the QueryElevationComponent.

Collapsing Query Parser Options

The CollapsingQParser accepts the following local params:

field

Required

Default: none

The field that is being collapsed on. The field must be a single-valued String, Int or Float-type of field.

min or max

Optional

Default: none

Selects the group head document for each group based on which document has the minimum or maximum value of the specified numeric field or function query.

At most only one of the min, max, or sort (see below) parameters may be specified.

If none are specified, the group head document of each group will be selected based on the highest scoring document in that group.

sort

Optional

Default: none

Selects the group head document for each group based on which document comes first according to the specified sort string.

At most only one of the min, max, (see above) or sort parameters may be specified.

If none are specified, the group head document of each group will be selected based on the highest scoring document in that group.

nullPolicy

Optional

Default: ignore

There are three available null policies:

  • ignore: removes documents with a null value in the collapse field.

  • expand: treats each document with a null value in the collapse field as a separate group.

  • collapse: collapses all documents with a null value into a single group using either highest score, or minimum/maximum.

hint

Optional

Default: none

There are two hint options available:

  • top_fc: This stands for top level FieldCache.

    The top_fc hint is only available when collapsing on String fields. top_fc usually provides the best query time speed but takes the longest to warm on startup or following a commit. top_fc will also result in having the collapsed field cached in memory twice if it’s used for faceting or sorting. For very high cardinality (high distinct count) fields, top_fc may not fare so well.

  • block: This indicates that the field being collapsed on is suitable for the optimized Block Collapsing logic described below.

size

Optional

Default: 100000

Sets the initial size of the collapse data structures when collapsing on a numeric field only.

The data structures used for collapsing grow dynamically when collapsing on numeric fields. Setting the size above the number of results expected in the result set will eliminate the resizing cost.

collectElevatedDocsWhenCollapsing

Optional

Default: true

In combination with the Collapsing Query Parser all elevated docs are visible at the beginning of the result set. If this parameter is false, only the representative is visible if the elevated docs has the same collapse key.

Sample Usage Syntax

Collapse on group_field selecting the document in each group with the highest scoring document:

fq={!collapse field=group_field}

Collapse on group_field selecting the document in each group with the minimum value of numeric_field:

fq={!collapse field=group_field min=numeric_field}

Collapse on group_field selecting the document in each group with the maximum value of numeric_field:

fq={!collapse field=group_field max=numeric_field}

Collapse on group_field selecting the document in each group with the maximum value of a function. Note that the cscore() function can be used with the min/max options to use the score of the current document being collapsed.

fq={!collapse field=group_field max=sum(cscore(),numeric_field)}

Collapse on group_field with a null policy so that all docs that do not have a value in the group_field will be treated as a single group. For each group, the selected document will be based first on a numeric_field, but ties will be broken by score:

fq={!collapse field=group_field nullPolicy=collapse sort='numeric_field asc, score desc'}

Collapse on group_field with a hint to use the top level field cache:

fq={!collapse field=group_field hint=top_fc}

Collapse with custom cost which defaults to 100

fq={!collapse cost=1000 field=group_field}

Block Collapsing

When collapsing on the _root_ field, using nullPolicy=expand or nullPolicy=ignore, the Collapsing Query Parser can take advantage of the fact that all docs with identical field values are adjacent to each other in the index in a single "block" of nested documents. This allows the collapsing logic to be much faster and more memory efficient.

The default collapsing logic must keep track of all group head documents — for all groups encountered so far — until it has evaluated all documents, because each document it considers may become the new group head of any group.

When collapsing on the _root_ field however, the logic knows that as it scans over the index, it will never encounter any new documents in a group that it has previously processed.

This more efficient logic can also be used with other collapseField values, via the hint=block local param. This can be useful when you have deeply nested documents and you’d like to collapse on a field that does not contain identical values for all documents with a common _root_ but is a unique and identical value for sets of contiguous documents under a common _root_. For example: searching for "grand child" documents and collapsing on a field that is unique per "child document"

Specifying hint=block when collapsing on a field that is not unique per contiguous block of documents is not supported and may fail in unexpected ways; including the possibility of silently returning incorrect results.

The implementation does not offer any safeguards against misuse on an unsupported field, since doing so would require the same group level tracking as the non-Block collapsing implementation — defeating the purpose of this optimization.

Expand Component

The ExpandComponent can be used to expand the groups that were collapsed by the CollapsingQParserPlugin.

Example usage with the CollapsingQParserPlugin:

q=foo&fq={!collapse field=ISBN}

In the query above, the CollapsingQParserPlugin will collapse the search results on the ISBN field. The main search results will contain the highest ranking document from each book.

The ExpandComponent can now be used to expand the results so you can see the documents grouped by ISBN. For example:

q=foo&fq={!collapse field=ISBN}&expand=true

When used with CollapsingQParserPlugin and there are multiple collapse groups, the field is chosen from the group with the least cost. If there are multiple collapse groups with same cost then the first specified one is chosen.

When enabled, the ExpandComponent adds a new section to the search output labeled expanded.

Inside the expanded section there is a map with each group head pointing to the expanded documents that are within the group. As applications iterate the main collapsed result set, they can access the expanded map to retrieve the expanded groups.

The ExpandComponent has the following parameters:

expand

Required

Default: none

When true, the ExpandComponent is enabled.

expand.field

Optional

Default: none

Field on which expand documents need to be populated. When expand=true, either this parameter needs to be specified or should be used with CollapsingQParserPlugin. When both are specified, this parameter is given higher priority.

expand.sort

Optional

Default: score desc

Orders the documents within the expanded groups.

expand.rows

Optional

Default: 5

The number of rows to display in each group.

When expand.rows=0, only the number of documents found for each expanded value is returned. Hence, scores won’t be computed even if requested and maxScore is set to 0.
expand.q

Optional

Default: none

Overrides the main query (q), determines which documents to include in the main group. The default is to use the main query.

expand.fq

Optional

Default: none

Overrides main filter queries (fq), determines which documents to include in the main group. The default is to use the main filter queries.

expand.nullGroup

Optional

Default: false

Indicates if an expanded group can be returned containing documents with no value in the expanded field. This option only enables support for returning a "null" expanded group. As with all expanded groups, it will only exist if the main group includes corresponding documents for it to expand (via collapse using either nullPolicy=collapse or nullPolicy=expand; or via expand.q) and documents are found that belong in this expanded group.