FR, merge datatourisme - Githubissues

frodrigo commented 4 years ago

cc @nlehuby @vinber Issue to improve datatourisme merge

First results with tourism_office http://osmose.openstreetmap.fr/en/errors/?item=8420&useDevItem=all

frodrigo commented 4 years ago

We should add the postal address in the subtitle.

nlehuby commented 4 years ago

I've started a branch: https://github.com/osm-fr/osmose-backend/compare/tourism_office

frodrigo commented 4 years ago

From place from data.gouv, 202062 lines Categories stats: https://pastebin.com/QtWqu7D0

All have PlaceOfInterest and PointOfInterest.

I think we can ignore all generic Accommodation and Lodging.

Low frequency contains interesting thing... but there have low frequency.

Camping or Church can be the next step as they can be found on imagery.

frodrigo commented 4 years ago

I make the analyser generic. https://github.com/osm-fr/osmose-backend/pull/883

nlehuby commented 4 years ago

Can we have something like this and let the user choose the kind of amenity/shop when we have several ones in the source dataset ?

frodrigo commented 4 years ago

Multiple fixes are not implemented in merge analyser. Should require to have multiple Generate parts. On Sirene is just done like this https://github.com/osm-fr/osmose-backend/blob/master/merge_data/shop_FR.mapping.json#L17-L18

frodrigo commented 4 years ago

Merge Restaurant, Camping and Church. Remove pervious Restaurant for Aquitaine and Saône-et-Loire

frodrigo commented 4 years ago

Also disable merge_tourism_FR_aquitaine_information.

We have analyser_merge_tourism_FR_aquitaine_camp_caravan.py but with:

tourism": "caravan_site
tourism": "camp_site

Can we have both from Datatourisme?

frodrigo commented 4 years ago

Camping http://osmose.openstreetmap.fr/fr/errors/?item=8140&class=330

Church, few, probably only touristic ones (hidden item) http://osmose.openstreetmap.fr/fr/errors/?item=8101&class=320&useDevItem=all

Restaurant http://osmose.openstreetmap.fr/fr/errors/?item=8240&class=340

We have a static conflation distance, we should move it into the json file.

frodrigo commented 4 years ago

I have setup a configurable conflation distance.

@vinber please, can you look about the ref? The in Datatourisme seam to be just the same as from local producer.

frodrigo commented 4 years ago

Camping have a wide range of application from farm camping to tourism=hotel.

vinber commented 4 years ago

yes, i think in case of TourinFrance (tourinsoft, sirtaqui for exemple ?) / the ref seems to be same than the local producer ([rdfs:isDefinedBy])

page 13 https://framagit.org/datatourisme/ontology/-/blob/master/Documentation/Ontologie%20DATAtourisme%20v2.4_Documentation.pdf

vinber commented 4 years ago

once again i'm lost :) where can i see datatourism data than you merge with osm ?

frodrigo commented 4 years ago

So what ref tag to use?

@vinber it is generated from a query https://github.com/osm-fr/osmose-backend/blob/master/analysers/analyser_merge_tourism_FR.py#L47 https://github.com/osm-fr/osmose-backend/blob/master/analysers/analyser_merge_tourism_FR.py#L70

vinber commented 4 years ago

good question, when we use it in Aquitaine, it was ref:FR:CRTA, but now with data tourism i don't know. I ask on mailing list ? ref:FR:tourism ?

frodrigo commented 4 years ago

But it is not a datatourisme ref. If it is just a copy of the local ref we may need to use the local ref key?

frodrigo commented 4 years ago

The adjustment of conflation distance double the number of Restaurant and Church.

vinber commented 4 years ago

But it is not a datatourisme ref. If it is just a copy of the local ref we may need to use the local ref key?

yes you're right

vinber commented 4 years ago

In datatourism, SIT (Système d'information touristique : local source) is define by hasBeenCreatedBy or hasBeenPublishedBy And ref is : isDefinedBy

vinber commented 4 years ago

and if there's no "possible_merge" but "only" missing_official it's because you have not time to do or for an other reason ?

frodrigo commented 4 years ago

and if there's no "possible_merge" but "only" missing_official it's because you have not time to do or for an other reason ?

We need a ref

vinber commented 4 years ago

I thought it was just for update_official

frodrigo commented 4 years ago

hasBeenCreatedBy

https://data.datatourisme.gouv.fr/864cf6ad-2809-39bf-a1aa-fbf3478cdf29

hasBeenPublishedBy

https://data.datatourisme.gouv.fr/7def446d-ec1e-327f-ba4e-262070d9785e

And ref is : isDefinedBy

For me the ref is under dc:identifier, isDefinedBy is aproperty of what?

Update query with ?creator_name and ?identifier https://annuel.framapad.org/p/dataljfjFKJQLKQLQ

vinber commented 4 years ago

page 9 https://framagit.org/datatourisme/ontology/-/blob/master/Documentation/Ontologie%20DATAtourisme%20v2.4_Documentation.pdf isDefinedBy : lorsque le concept est issu d’un thésaurus déjà existant. Renseigner ici l’identifiant unique du terme concerné.

But i'm not sure to understand what i read.

vinber commented 4 years ago

I suppose you're right : "La référence du POI dans le SIT source est dans la propriété dublin core identifier La référence unique du POI dans DATAtourisme est son URI"

https://support.datatourisme.gouv.fr/t/ref-du-sit-regional/1086/2

frodrigo commented 4 years ago

"concept" are the meta model, not the data, hasBeenCreatedBy or hasBeenPublishedBy are concept.

frodrigo commented 4 years ago

The issue is now that Osmose merge framework does not support variable tags key (but only variable value). We can filter data and run only on data from a source by area, or try to make support of variable key in the framework.

frodrigo commented 4 years ago

I change query to use ref (@nlehuby I switch the query to my own datatourism account) I map the ref from SIRTAQUI Nouvelle-Aquitaine" to ref:FR:CRTA, like disabled tourism query. @vinber I list the available provider at the end of https://annuel.framapad.org/p/dataljfjFKJQLKQLQ

frodrigo commented 4 years ago

frodrigo commented 4 years ago

Major classes integrated and activated on frontend. Thank all.

osm-fr / osmose-backend

FR, merge datatourisme #875

Nouveaux

Déjà présent

Tout