This feature can help improve our machine learning algorithm
Some of the 51Degrees services benefit from and are greatly improved by optional usage sharing from customers, where data (or “evidence”) is sent to the 51Degrees data processing system. This evidence could be information in a HTTP header like a User-Agent string or User-Agent Client Hints.
While this feature is optional for our on-premise users, we encourage you to enable usage sharing so that our algorithm can discover new or rare devices, thus improving the accuracy of our Device Detection service.
Background to usage sharing
On-premise users can configure the installation to disable usage sharing. Users of our cloud service cannot disable usage sharing; for us to detect the device, we require the User-Agent/User-Agent Client Hints evidence. To find out more, please refer to our usage sharing feature guide.
Our usage sharing algorithm can quickly decipher usage data that is sent to it. On the client, this can monitor repeated evidence to preserve memory and bandwidth. Additionally, the usage sharing algorithm can disregard shared evidence that does not contain the necessary information server-side. In high-traffic scenarios, the algorithm is also able to sample the data by receiving a certain percentage of evidence shared. This means that unnecessary data not used by the algorithm is not sent, and we don’t consume too much of the clients' resources.
To ensure your data is received by our algorithm, we recommend reviewing your configuration of the 51Degrees service to ensure:
Usage sharing is enabled.
You provide data for all the required evidence fields.
The algorithm can only receive evidence if the required fields contain evidence. The required fields are:
User-Agent (header.user-agent or query.user-agent)
Client IP Address (server.client-ip)
What do we do with the evidence data sent via usage sharing?
Certain evidence fields are only required to verify that the data we receive is legitimate. We only collect Client IP Address evidence so we can ensure the data is widely represented by what is seen in the real world. For this representation to be accurate, we require non-obfuscated and correct Client IP Address data. We do not use the Client IP Address information for anything other than legitimacy-validation.
You can also optionally supply evidence for the following fields:
Website (header.host, header.usage-from, or both)
User-Agent Client Hints (sec-ch-ua, sec-ch-ua-full-version-list, sec-ch-ua-platform, etc.)
We use the information provided by usage sharing, such as the User-Agent and User-Agent Client Hints HTTP headers, to train a separate machine learning algorithm to learn new devices. Additionally, we use the header.host and header.usage-from fields to trace the usage from a particular website or customer, helping us to investigate should they raise a query.
To learn more on how to provide evidence for the User-Agent Client Hints (UA-CH) headers, please refer to our UA-CH documentation.
The terms for your current pricing plan (Free, Big, Bigger, or Biggest) contain more information on how we use the data you send us. Additionally, the data addendum section has more information on our use of your data within specific privacy laws, such as the Data Protection Act 2018 and the General Data Protection Regulation (GDPR). As ever, if you have any questions, please do not hesitate to get in touch.