> For the complete documentation index, see [llms.txt](https://adityamukho.gitbook.io/recallgraph-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://adityamukho.gitbook.io/recallgraph-docs/understanding-recallgraph/terminology/grouping.md).

# Grouping

Many bulk read endpoints support an optional grouping parameter, using which results can be grouped together on some attribute. Common grouping attributes include:

1. **Node:** Group by the `_id` of the node (vertex or edge) represented by an entity ([event](/recallgraph-docs/understanding-recallgraph/terminology.md#event), [command](/recallgraph-docs/understanding-recallgraph/terminology.md#command) or historical version) in the result set.
2. **Collection:** Group by the *collection* of the node represented by an entity in the result set.
3. **Type:** Group by the *type* (vertex or edge) of node represented by an entity in the result set.
4. **Event:** Group by the *type* (created/updated/deleted) of event that the entity in a result set represents.

{% hint style="info" %}
It is not necessary that all of the above grouping attributes are available for all endpoints. It depends on the core structure of the data requested.
{% endhint %}

If a grouping attribute is specified, an additional set of sub-parameters can be optionally specified that determine how the entries within a group are arranged. These are:

1. **Group Sort:** Sorts entries within a group by a specific attribute. The attribute choice depends on the specific endpoint. See API docs for more details.
2. **Group Slice:** Skip and/or limit entries within a group, post group-sort.
3. **Aggregation (Count):** Rather than return an entity list within a group, return their count (group-wise totals). If this sub-parameter is specified, the other sub-parameters are ignored.

{% hint style="info" %}
There is a special case of aggregation, wherein counts are requested, but the grouping attribute is absent. In this case, the API returns the total count of ALL entities, rather than group-wise totals. In this case, the upper level sorting and slicing parameters are ignored.
{% endhint %}

{% hint style="info" %}
Grouping clauses are applied within the DB query that is used to fetch intermediate results.
{% endhint %}