include facet stats in a search result

[pvgenuchten]

This is probably relevant for oapi-common, but i want to raise it here first, because the faceted search concept is currently most common in the catalogue domain, but i can see similar uses in oapif & sensor api's.

A typical use case in catalogues is faceted search to further limit a search result. Faceted search requires an analyses of the search result best managed during the backend query, resulting in some statistics being added to a search result.

What would be the best approach/location in a typical OAPIR search response to include this information and does it make sense to suggest this as a common extension to OAPI?

[mhogeweg]

I see facet information as a property of the catalog as a whole (or of a collection), not so much specific to a search. when performing a search, selections made in facets translate to a search expression that could also have been generated based on a different UI or based on CQL/Filter style queries.

[pvgenuchten]

my request relates to facet information specific to the search result, so you query for "forest", then the search result will return a first page of results plus facet statistics on the total resultset (how many maps/services/datamodels/datasets, how many results in projection a/b/c, how many results in language eng/fre/ger, etc). See for example the geoss portal

for relevant protocols/keywords is per search result indicated how many results in the current result set

[mhogeweg]

ok, I think that would make for a nice extension. It would require some discussion as not all facets are as clean as a term+count. In Geoportal Server we support temporal, numeric, and spatial facets (see http://geoss.esri.com/geoportal_264):

We see faceted search as a 180 degree turn from the traditional 'google one box' search experience. Rather than guessing terms to look for something, you start with all the catalog has to offer. Through the choices made in the facets, you discard what is not relevant, whether it be spatially outside your area of interest, or temporally not the right period, or not the right service type, etc etc. Every choice made in one facet also filters content in the other facets, always leaving you with what matches all of the choices. You can quickly broaden the view by removing a facet choice.

Different catalogs (collections?) may include different facets, suitable for the content in the catalog. for example, same technology, now showing information from a series of Spatio-Temporal Asset Catalogs: http://geoss.esri.com/imagecatalog/

Is the goal of including this in the API to build the UI on top of that?

[pvretano]

2020-11-02: This capability is a nice-to-have but it is not something that would be in the core but rather would be an extension.

[pvgenuchten]

Sure, let's discuss it in the scope of an extension

[tomkralidis]

FYI #372 introduces the setup for OGC API - Records - Part 2: Facets. Perhaps we can work on this at the July sprint @pvgenuchten @fgravin @jahow?