keyword searching

[antje-s]

In the keyword search it is apparently possible to search for several keyword values, whereby the default operator ‘AND’ seems to be used, e.g.

1. prediction
   https://wis2.dwd.de/gdc/collections/wis2-discovery-metadata/items?f=json&limit=200&keywords=prediction
   --> numberMatched = 58
2. prediction,global
   https://wis2.dwd.de/gdc/collections/wis2-discovery-metadata/items?f=json&limit=200&keywords=prediction,global
   --> numberMatched = 34
3. prediction,global,"Weather and Climate"
   https://wis2.dwd.de/gdc/collections/wis2-discovery-metadata/items?f=json&limit=200&keywords=prediction,global,%22Weather%20and%20Climate%22
   --> numberMatched = 29

It seems to be possible to change the operator to OR, e.g. via
https://wis2.dwd.de/gdc/collections/wis2-discovery-metadata/items?f=json&limit=200&q=(keywords:prediction%20global)
--> numberMatched = 96

Does this behaviour apply in principle?

[kaiwirt]

Push :-)

[tomkralidis]

Does this behaviour apply in principle?

Similar to #25, for single term equality queries, I would suggest a "safer" way to query with:

https://wis2.dwd.de/gdc/collections/wis2-discovery-metadata/items?f=json&limit=200&keywords=prediction%20global

For compound queries with specific syntaxes such as AND OR NOT, etc., I would delegate to your examples of q per above. These examples are also in the WIS2 Cookbook (please feel free to suggest more examples there as well).

[kaiwirt]

A search for
https://wis2.dwd.de/gdc/collections/wis2-discovery-metadata/items?f=json&limit=200&keywords=%22Quantity%20of%20precipitation%22|%22Analysis%22
returns these
"keywords":[
"Historical",
"Archive",
"Hydrology",
"Hydrometric",
"Streamflow",
"Discharge",
"Water level",
"Water quantity",
"Water quantity",
"Sediment concentration",
"Water",
"National (CA)",
"Observation",
"Measurement"
],
What would be the best way to search for "Quantity of precipitation" OR "Analysis"
What we would like to have is a search for the exact terms (in ") combined with OR

[tomkralidis]

@kaiwirt you could use the q parameter which triggers ES query syntax:

https://wis2-gdc.weather.gc.ca/collections/wis2-discovery-metadata/items?f=json&q=properties.keywords:%22Quantity%20of%20precipitation%22%OR%20properties.keywords:Analysis

[joergklausen]

It's the example "Quantity of precipitation" that triggers me here (admittedly, a bit out of context): Please try to support a more unified vocabulary approach based on the WMO/WIGOS code lists, which in this case, would use "Amount of precipitation".

[tomkralidis]

Thanks @joergklausen. WCMP2's properties.keywords is designed to be free form keywords and does not have to align to any controlled vocabulary (1).

WCMP2 provides themes.concepts, allowing for 1..n concepts qualified by a scheme or register/dictionary) (2). Having said this, in WCMP2 we do need to add further checks/KPIs to ensure validation of these terms against their respective CVs.