What's new in release 20220411

Added location fields generated_intro and generated_snippet

Similarly to what we have recently done for POIs, we extended the location descriptions fields adding generated_intro and generated_snippet which, as their names suggest, correspond to the intro and snippet fields but with the advantage of being 100% attribution-free because they are entirely generated using our algorithms.

Improved eatingout and nightlife POI generated descriptions

Recently we released new fields for POIs, called generated_intro and generated_snippet. We've constantly put effort in making them better and better and with this version we think we reached very good quality especially for eatingout and nightlife POIs.

Improved artist and architects POI connections

When searching for tag_labels for a POI you might find some related to specific artists and architects. This was true even before this new API version, but we were mentioning only some which adhered to very strict rules and now we extended the logic to include many more of them that have a relevant connection to a POI. For example, for the Louvre Museum (i.e. with id T__34e2331607b3) you could have found Monet as an artist but in this API version you'll find 24 more!!

That's not all! Under the hood this implies that artists and architects get connected to other locations as well. Keeping Monet as an example, you'll see that his tag (i.e. with label person_artist-claude_monet) won't be found only in French locations but in many others as well, for example Vienna and New York City!

Updated and improved data

As usual, we freshened data coming from our sources and we put effort into improving the algorithms which use them in order to generate better quality outputs. Amongst these improvements we want to mention the improvement of the tags score normalization algorithm which fixes a few cases where the final tag score was higher than 10.

What's new in release 20220104

Improved and Updated data

As always, this new release provides updated data coming from our sources and includes various small improvements to our algorithms.

What's new in release 20211011

Added poi fields generated_intro and generated_snippet

In all versions we have three types of POI descriptions:

  • structured_content for the complete description
  • intro for medium-length sentences
  • snippet for shorter ones

Whenever it's possible we try to provide descriptions which sound as natural as possible by using text extracted from our authoritative sources. But unfortunately that's not always possible for several reasons, therefore in these cases we generate our own sentences using POI features. One other thing to consider is that the descriptions not built by us need source attribution for usage and that can be a burden.

For this reason we've added two new fields generated_intro and generated_snippet which, as their names suggest, correspond to the last two fields mentioned above but with the peculiarity of being 100% attribution-free because they are entirely generated using our algorithms.

Updated data

Freshened data coming from our sources and updated all of our content accordingly.

What's new in release 20210615

Parking data

In this release, we have added more than one million parking POIs globally. To find a parking, simply look for pois with the parking tag label. We have also added two new tag labels relevant to parking. In more detail:

  • parking (improved coverage)
  • subtype-Paid_parking (new tag label)
  • subtype-Customer_parking (new tag label)

See the reference documentation for more information about the available tag labels.

This way, one can exclude some paid and/or customer-only parking options when looking for a parking. Note that the absence of subtype-Paid_parking tag label, does not guarantee that the parking is free. For more details on how to look for parkings, take a look at the relevant examples. It is now easy to find out where to park your car during your adventures!

Improved best_for tag scoring algorithm

The best_for tag is used to mark a POI which is exceptionally good for something in a location. In this version, we have improved the tag's assignment and scoring process.

Improved location ids for some towns

In previous versions, some locations would get their ids as a derivative of the name of the municipality they belong to. In this version, we prefer to assign the id based on the name of the town and not the relevant municipality. This change affects mostly locations in Belgium.

Reject boring locations

We are dropping locations that only have POIs of low interest for a tourist. "Boring" POIs are considered to be bus stops, ATMs, police stations etc. On the other hand, POIs that are considered interesting for a tourist include restaurants, bars, cinemas, museums etc. The "boring" POIs are not lost though. Instead, they are retrievable via a nearby or a parent location. This way, you can still find something that you might need during your trip, such as an ATM, in a short distance from where you are.

Updated data

As always, we have update our sources to include more up-to-date details!

What's new in release 20210317

Improved data

We have added an extra layer to the region hierarchy for the following five countries: USA, UK, Spain, Italy and China. This means that from now on we will have regions like: Wales, England, Scotland, New England, Midwest, etc.

We also improved the assignment of pois to locations, especially for overseas territories and for locations that are part of larger locations (e.g. Hollywood, which is also part of Los Angeles).

As usual, this release contains many other small improvements to our algorithms as well as refreshed data.

Articles for cuisine tags

We have added articles to our cuisine tags when available.

Deprecated fields removed

This change is backwards incompatible

In this release we have removed a number of fields that have been deprecated for a long time. In particular, we have removed all the fields of the old type content, which was deprecated in favour of structured_content. The affected types are location, poi, tag, tour and article. We've also removed the vendor and vendor_tour_url fields from the tour type, which were deprecated in favour of the booking_info field.

Note that this change does not affect older API versions.

What's new in release 20201111

Subfields support

Previously, if you wished to specify which fields to include in an API response, then (with some exceptions) it was only possible to do so for the top-level objects returned; any nested objects always had their default set of fields, with no way to change it. It is now possible to specify which fields to return for any given type for any endpoint. See the documentation for more details.

Note that this feature has been backported to all supported API versions.

Deprecated fields not included by default

Now that it is possible to specify which fields to return for nested types, we no longer include by default any fields that are deprecated (this affects the article, image and tour types).

Improved and updated data

We have improved our data, adding better support for overseas departments and territories. As part of this, many such territories are now treated as top-level locations. We also put some effort in getting more national parks and nature reserves and improving the algorithms that decide which POIs should be assigned to them.

As usual, this release contains many other small improvements to our algorithms as well as refreshed data.

What's new in release 20200803

New or improved tags

We have added a number of new tag labels and improved the coverage of others, in particular:

  • feature-Delivery (new tag label)
  • feature-Drive_through (new tag label)
  • feature-Takeaway (new tag label)
  • feature-Outdoor_seating (improved coverage)
  • character-Chain (new tag label)

See the reference documentation for more information about the available tag labels.

Updated data

As always, this release contains refreshed data and various fixes and improvements to our algorithms.

Merging of our enterprise and non-enterprise releases

With this release, we no longer have separate data versions for our enterprise customers. The biggest impact for our non-enterprise customers is that we no longer include content (text or images) from Facebook. For our enterprise customers, combining the releases means a significant increase in the number of objects in the database (e.g. 13% more POIs than the last enterprise release, including nearly 50% more restaurants).

Location ID changes

In general, we try to keep IDs as stable as practical between data releases. However, in order to solve some issues and improve things going forward, we've made changes that mean that you might notice a higher rate of change in location IDs in this release than usual (but still small).