This repository is now archived and subsequent works will be carried out in Smart data models repositories.
This repository contained:
- JSON Schemas and documentation on harmonized datamodels for different Smart Domains, particularly Smart Cities and Smart Agrifood.
- code that allows to expose different harmonized datasets useful for different applications. Such datasets are currently exposed through the FIWARE NGSI version 2 and/or NGSI-LD APIs (query).
This work was aligned with the results of the GSMA IoT Big Data Project. Such project is working on the harmonization of APIs and data models for fueling IoT and Big Data Ecosystems. In fact the FIWARE data models are a superset of the GSMA Data Models.
Some of the Data Models in this Repository have been adopted by a joint initiative between TM Forum and FIWARE Foundation. For more info please check https://github.com/smart-data-models/dataModels
📚 Documentation |
---|
To support the adoption, we created a short guideline for the usage of data models. If you are using NGSI-LD, you should also check the NGSI-LD HowTo and the NGSI-LD FAQ.
A JSON Schema was provided for every harmonized data model. There are different online JSON Schema Validators, for instance: http://jsonschemalint.com/. For the development of these schemas the AJV JSON Schema Validator is being used. For using it just install it through npm:
npm install ajv
npm install ajv-cli
A validate.sh
script is provided for convenience.
Note: JSON Schemas capture the name and data type of each Entity Attribute.
For instance, this means that to test JSON schema examples with a
FIWARE NGSI version 2 or
NGSI-LD
API implementation, you need to use the keyValues
mode (options=keyValues
).
See:
- https://gitlab.com/synchronicity-iot/synchronicity-data-models
- schema.org
- https://github.com/GSMADeveloper/NGSI-LD-Entities
- https://forge.etsi.org/gitlab/NGSI-LD/NGSI-LD
MIT © 2019-2021 FIWARE Foundation e.V.
All the code in this repository is licensed under the MIT License. However each original data source may have a different license. So before using harmonized data please check carefully each data license.
All the data models documented here are offered under a Creative Commons by Attribution 4.0 License.