Back to discover
NewsBuild with AI

Vespa adds a new search path for product properties

Fast-search map fields target filtering across varied product catalogues, pairing property names with their values.

Conceptual illustration of miniature products on trays with paired blank tokens; an amber frame highlights the pair beside a blue bicycle.
CurioFuse · AI-generated illustration

Vespa has introduced fast-search map fields for catalogues whose products have different attributes. A map stores property names with values, accommodating details such as bicycle frame size or phone battery capacity. According to the announcement, the new approach addresses a costly case: when many documents share a property name, the existing search must still check which values belong to it.

The schema documentation makes the adoption requirements explicit. Developers add a synthetic search field to the existing map, but must re-index before that field becomes searchable. This also changes how data can be updated: assigning or removing an individual map key through a field path is disabled. To change existing entries, the entire field must be assigned. For arrays of structured entries, updating or removing one element is likewise disabled, while adding elements remains supported.

The query reference describes how to address the new field. The mapMatch operator takes two arguments: one condition for the key and another for its value. String conditions use contains; numeric values allow numeric comparisons or ranges. Queries use the full name of the synthetic field. The documented price example therefore supports either an exact price or a price range associated with that key. A shorter bracket notation expresses the same searches.

Sources(3)

Sources & standardsReport a correction