Reference Entry
What Querylight TS Can Do Beyond Full-Text Search
Overview · foundation · order 62
Querylight TS can power more than lexical retrieval, including faceting, aggregations, semantic retrieval, geo filtering, and lightweight dashboards.
Relevant APIs
What Querylight TS Can Do Beyond Full-Text Search
It is easy to look at Querylight TS and assume it is only a small browser full-text search library.
That would miss a large part of the design.
The library already supports several capability layers:
- lexical retrieval
- faceted filtering
- numeric and date aggregations
- significant terms
- vector search and rescoring
- geo indexing and geo queries
- serialized index state for build-time shipping
The new dashboard demo extends that story into lightweight local analytics.
Search is still the center
The search use case is still real and important:
- docs search
- static site search
- app help centers
- small embedded retrieval tasks
But once your documents include structured fields, the same indexes can support more than ranked text matching.
Examples of "beyond search" behavior
Faceted discovery
Build sidebars and active filters from:
termsAggregation(...)rangeAggregation(...)histogram(...)dateHistogram(...)
Lightweight local dashboards
Use:
TermQueryTermsQueryRangeQuery- numeric/date field stats and buckets
to derive chart-ready subsets from raw records.
Semantic retrieval
Use:
VectorFieldIndex- semantic embeddings
- reranking or nearest-neighbor search
for question answering or related-content experiences.
Geo-aware filtering
Use:
GeoFieldIndexGeoPointQueryGeoPolygonQuery
for location-based slices and map-adjacent interfaces.
Why this matters
For small browser and local-first systems, it is useful when one toolkit can cover several adjacent needs instead of forcing an early split into:
- one library for search
- one library for analytics
- one library for geo
- one extra backend just to make the UI interactive
Querylight TS does not replace every specialized system. But it can cover a wider practical surface than “text in, ranked hits out”.