Hotel mapping solutions for travel tech and AI companies

Travel tech and AI companies need clean, duplicate-free and structured hotel data for search, recommendation and AI models. hotelmapping.org identifies every hotel with a single HM ID, links any record to this ID with the Find Match API and delivers multilingual content via API and daily dump.

Problems you face

  1. 1

    Duplicate records mislead the model

    If the same hotel appears several times in the dataset, search results repeat, the recommendation system presents the same hotel as different options and the AI assistant gives contradictory information.

  2. 2

    Fields are free text

    Room name, board type and area arrive as free text from most sources. If "UAI", "Ultra AI" and "Ultra Her Şey Dahil" remain separate values, filtering and structured search do not work.

  3. 3

    Building your own mapping is expensive

    Normalization, transliteration, candidate generation, scoring, a labeled test set and human review require a separate team and continuous maintenance. This is not the core business of most product teams.

Recommended package

Find Match API + hotel mapping + content and dump

The Find Match API links any record to an HM ID; hotel mapping provides the duplicate-free reference list; hotel content gives multilingual descriptions and factsheets. Full data is retrieved via daily dump. AI-ready data with schema.org JSON-LD, embeddings and a semantic search API is planned for a later stage.

Sample flow

  1. 1

    Sandbox access

    You try the endpoints in the sandbox environment with an API key. Authentication is done with the X-API-Key header.

  2. 2

    Mapping your existing records

    You send your hotel records to the POST /v1/match endpoint with name, address and coordinates; an HM ID and confidence score come back for each record.

  3. 3

    Retrieving the reference data

    With GET /v1/dumps/latest, you download the daily full or delta dump as CSV, JSON or Parquet.

  4. 4

    Connecting to your product

    Your search index, recommendation system or your model's knowledge source uses the HM ID as the key; board and area fields arrive with standard codes.

Why does structured hotel data matter?

For a search engine, recommendation system or AI travel assistant to answer correctly about a hotel, it must first resolve the question “which hotel is this record?” If records of the same hotel from different sources have not been merged, the model recommends the same hotel several times or mixes up the information of two different hotels. For the foundation of the topic, see the what is hotel mapping guide.

A team building its own mapping pipeline has to develop and maintain normalization, transliteration for Russian and Arabic names, geographic candidate generation, a scoring model, a manually labeled test set and a human review screen separately. hotelmapping.org offers this work as a single service; you work only with the HM ID.

Find Match API example

The Find Match API links a record you hold to an HM ID. Request:

curl -X POST https://api.hotelmapping.org/v1/match \
  -H "X-API-Key: $HM_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "name": "Sunset Bay Rsrt. & Spa", "city": "Belek", "country": "TR", "lat": 36.8539, "lng": 31.0556 }'

Response:

{
  "decision": "auto_match",
  "hm_id": "HM-1048213",
  "name": "Sunset Bay Resort & Spa",
  "confidence": 0.984
}

Sample data; the hotel name and codes do not belong to a real property. The decision field shows the decision made from the score: auto_match, review or no_match. You can take only auto_match records directly into your index and leave review records to a separate review step.

What do you get?

Data Product or format Status
Persistent hotel identity Hotel mapping Available
Record-to-HM ID matching Find Match API Available
Multilingual description and factsheet Hotel content Available
Full and delta dump CSV, JSON, Parquet Available
JSON-LD, embeddings, semantic search AI-ready data Planned (later stage)

Update frequency

Static hotel data changes constantly: hotels open, close and change names. The daily delta dump brings only changed records; the full dump is used to rebuild the index from scratch. Webhooks for change notifications are planned. For details, see the static hotel data term.

We explain how Russian and Arabic names are matched under the transliteration term. All endpoints are listed on the developers page; when choosing a provider, you can also look at the 12 questions when choosing a hotel mapping API list.

Frequently asked questions

Can we use your data for model training?

Usage and license terms are set in the contract. We recommend stating your intended use at the quote stage.

In which languages is content available?

The first languages of hotel content are planned as TR, EN, DE, RU and AR. Russian and Arabic hotel names are also transliterated to Latin script during mapping.

When is AI-ready data coming?

schema.org Hotel JSON-LD, embeddings and a semantic search API are planned for a later stage. Today you can use structured data through the HM ID, standard codes and dump files.

Request a quote

Get your tailored quote

Send your request in three steps; we will email the quote to your work address. The request is non-binding.

  1. 1 Select solutions
  2. 2 Select suppliers
  3. 3 Leave your contact details