Enhance search by fetching multiple requests ?

[etienneJr]

To improve the photon search on cartes.app, I thought of a strategy that would involve running several queries in parallel and sorting the results.

For example (but I haven't precisely thought about it yet). :

• a standard unfiltered query

but also :

• a query filtered on layer=city (this is a very classic query, I understood the standard query already gives high weight to cities)
• a query filtered on osm_tag=shop
• etc

and also identify keywords in the search text to run even more specific filtered queries, for example :

• test if the search text includes hostel to run a query with a filter on tourism=hostel
  • but we can't run only this query since other POIs may have hostel in their names without being an hostel
• test if the search text includes street to run a query with a filter on layer=street
• etc

Then I imagined comparing and re-sorting the results according to their score (probably with additional weights), but I now realise that the api doesn't provide such a score, just the sorted results, right ?

So I have 2 questions :

• do you think this is a valid strategy ?
• how can I get the score of each result in each query ?

Thanks !

[laem]

Have you seen #32 (comment) ?

[lonvia]

You basically suggest to implement an entirely different search algorithm. If you want to go down that road, I'd rather suggest you play around with the query algorithms in Photon directly. You can probably build one complex OpenSearch query to cover all your cases.

[etienneJr]

Well, your advice makes perfect sense to me, but diving into photon source code was exactly what I wanted to avoid ... 😅​ I need to think about it.