Review Places First - Week #3, 2019!


#1

Hi All,

This is a first message from me after Welcome.

First of all, thank you very much for interest and support was already given to me, even though there is no official team yet but me & my colleagues from OsmAnd already put some effort to make another step.

The main message I would like to present that we need Import Places First. You would wonder, why do we need review places where our main goal is to make Reviews for Places first. Please let me explain some steps:

  • To create a review system in the app like OsmAnd we need to have this data merged with OSM
  • In order to merge data with OSM we need to have stable IDs or links
    • OSM doesn’t have permanent IDs and doesn’t plan to have them - read more.
  • To create a permanent id that will track an OSM objects, special database is needed

Of course, we could have imported all POI’s from OpenStreetMap and create a database though it won’t solve a problem of changing name, relocating businesses and connecting additional information to it which anyway will be a manual task. Sometimes one restaurant would change a name but that would mean totally different entity which would need different set of reviews.

So, we think to make the process of Place registration manual and combine with other very useful steps for the system, so Moderator could:

  • Verify that place exists
  • Connect an official website
  • Choose the most appropriate image from open services like Mapillary
  • Connect proprietary places

Let’s explain one by one why these steps are needed:

  • Verification that place exists will help to contribute back to OSM
  • Lots of OSM places doesn’t have website which is the greatest source of public information and it could be contributed directly to OSM. Though it is hard to do in the field while mapping, mappers often skip that bit of very important information for later, but it is very easy to fix by just searching Google or DuckDuckGo for an official website. Obviously Google as an indexer doesn’t have any copyright on the link to the official website, so we are free to use it.
  • Images help a lot for end-users and make application better. In OsmAnd we show all Mapillary images automatically, but obviously some pictures are better than others and if it is easy to select the best picture, it would definitely help apps to make a use of OSM and Open Services in a better way.
  • That might sound strange but there is a lot of sense to connect even proprietary sources. Obviously, we can’t use any information from them! If TripAdvisor, Booking.com provides a public link to a place, we still can link it and store in in database. One of the use case is to import reviews once the user will grant permission to share review with OpenGeoReviews . Reminder: copyright for reviews belongs to the Users not to the Services and if User grants permission to duplicate review on OpenGeoReviews there is no restriction to do that, unless User signed a special agreement transferring all the Rights to the Service.

All that data will be distributed under Open License, we are about to choose between CC-BY-SA 4.0 or ODBL 1.0.

Anyway I hope to have a vivid discussion how to organize it better or what fields are important to validate first. That’s why I think to share Design before it gets built.

All activities are organized by Tasks similar to MapRoulette challenges. Tasks will be crowdsourced and we plan to start with 1st Task “Register Places” which might be never ending. Obviously, there will be freedom to select a specific place or to filter Places in Tasks by Location or by Category, later it will be possible to filter by other fields, for example by Language.


Here you could see the progress or pick up a place from the map directly.


To Edit or Register a place the most important is to specify Sources where information is taken from. As of today we don’t want people to fill in information themselves about a place and only connect from different sources. In case you know the Truth on the Ground it is better to put that information in OSM directly and it will be updated automatically, but it is very important to connect all the sources.

Later on you could use sources to specify direct value in the field. In case of OSM it will not be modifiable cause you can directly modify it on OpenStreetMap.

Again we don’t want to use OpenGeoReviews as a primary source of information for now. That’s why specifying a Source is very important. It will also help to automate some imports later and keep the License sanity.

Right now we plan to start with POI Food category like restaurants, bars, cafes, … and later on keep adding other categories. Set of fields will be also limited on UI to name, phone, opening hours, address (obviously backend will have a generic tag/value scheme). Later on, once the Wiki component (similar to OSM) will be available the list of categories and fields will start growing.

Note: feel free to post other posts on Forum and comments here to have a live discussion. Also our Telegram channels are available for spontaneous discussions .

P.S. We are eagerly looking for any (backend, frontend) developers who can help to build the system, please feel free to reach me :slight_smile:

Thanks for reading!
Victor


#2

That’s very nice. Unfortunately I am not a developer, sorry :confused: I really hope this project takes off.


#3

No worries, please think that you can contribute to the data as a moderator. We need a small team for development 1-5 people but huge community 100-500 at least. So, it is more important if you agree with ideas and ready to contribute once we build the first editor/


#4

You appear to be further ahead than I expected!
Looks really good. As mentioned on telegram, I would’ve found direct osm editing from the import screen less cumbersome


#5

I actually was thinking other way around, it is better to integrate places first and then we could think of automatic plugin for JOSM to cross-validate the changes. We also need to be careful in a first place and don’t rush to edit OSM to double check License and local community emotions.


#6

That sounds like a pretty good start! :+1:

But I’m not sure to understand why we need to have an Edit action: we “only” need to create a database with a stable id for each place, and a set of link / id to other sources, right ?

If we can edit (by choosing the value of fields from a source), we are creating a concurrent places database.
And if the output of this “Import Places” community task is a places database released under an open license, it will be used as a primary source of information and there would be no real virtuous circle: this database would be more complete than OSM (because it would contains more fields / “tags”, taken from our other sources) and I would have no incentive to use OSM instead of this database or to contribute to OSM if the other sources already contains the fields I’m interested in.

I like the idea to combine sources to enhance place information and complete fields that are missing in OSM, but wouldn’t it be better if this community task only suggests tags editions directly in OSM ?


#7

Indeed it is a valid concern to have 2 databases serving 2 purposes but it could be answered. The issue we couldn’t succeed with complete Places DB in OSM cause it is very big project with multiple goals and most important it is focused on mapping and geometry where Places DB should be focused on details of the place. There is no way we could put in OSM DB pictures or reviews and it shouldn’t be there to be honest. I followed quite many discussions where OSM should change to be able to get more data under different conditions but I think OSM shouldn’t change and instead it should have lots of satellite projects like Mapillary etc.

Please note, I actually suggest to combine information under the compatible license for OSM which later could be imported without a problem cause it will have 2 major advantages: 1 to 1 mapping to OSM ids and clear specification of source of information.


#8

Could wikidata be this other database ?

It has unique & stable id and a compatible license
It is already kind of linked to OSM as we can already add wikidata id as OSM tag
It can have identifiers to other helpful databases

See for instance this restaurant in OSM : https://www.openstreetmap.org/node/2193015893
We have in OSM all useful detail about the place and a wikidata id : Q60615583

Wikidata gives us

  • images (via Wikimedia Commons)
  • yelp id
  • trip advisor id
  • facebook id
  • Gault et Millau id

#9

Sure, as I see this example of Wikidata shows exactly how we need to structure information as well, though I believe we need to bring it to another scale comparing how many Restaurants today in Wikidata (2726) and our target is at least 2 000 000. I think later that data could be contributed back to Wikidata or leave in a separate project.