The objective is to prefetch that information for a list of entities without having to lookup each single property table to match the entity. This means that for a list of 500 result matches we only require one SQL query to know which properties and categories are used on all 500 entities including they amount of values assigned to each property.
The information is stored in a simple blob field type since we do not want to query the information in the field. That information is stored separately. Rather we only want to fetch the entire content at once to build an aggregation map.
- Semantic MediaWiki: GitHub pull request gh:smw:4613