Machine Learning for User and History Based Recommendations

For Netmedia Europe, Oakwood has built a tailor-made recommender system supported by Machine Learning-techniques. Training data for the machine learning model consisted of millions of interactions of their users on their platforms.

The solution we have provided for them is end-to-end: data is extracted from their AWS S3 buckets in real-time and the machine learning model is updated in real-time to provide low-latency recommendations to their users.

Big data

Netmedia has built up a large scale dataset of millions of interactions of their platform users. This pool of data is a valuable resource to mine for insights, however it can be quite a challenge to work with such a huge amount of data. By setting up a well-defined and optimized data pipeline, even big data is only a small job for the recommender solution we have provided Netmedia with.

Real-time (re)training

Users are continuously assigned to generalized profiles by the machine learning model, which takes into account past user browsing history and behavior on the platform.

Multi-level layered recommendations

Recommendations are performed at multiple different levels at the same time, being the brand, retailer and different product category levels. This allows Netmedia to provide different levels of recommendations to their platform users.

Low-latency prediction times

Netmedia Europe’s primary requirement was that predictions should happen in a matter of milliseconds, as this heavily reflects on their rankings in search engines which are an important source of traffic for their platforms. We have solved this challenge by using the latest and greatest technologies optimized for both speed and accuracy.

Top notch performance with industry standards

The entire solution can be fully integrated in their workflow to even better serve their platform users. A REST-compliant API is exposed to which a user’s most recent history and/or unique identification can be sent to, which in turn returns recommendations tailored for the specific user in an industry standard and efficient way.

Would you like to learn more about the different recommendation engines we can offer you and how they integrate with your technologies? Feel free to contact us.

To find out more about Netmedia-Europe and the services they offer, check out their website on


[Total: 2    Average: 5/5]
By | 2019-04-01T17:39:28+01:00 September 5th, 2018|Categories: Artificial Intelligence, Automation, Machine Learning, Recommender Systems, Use cases|Comments Off on Machine Learning for User and History Based Recommendations

About the Author:

Koen is co-founder of Oakwood and specializes in Artificial Intelligence, Machine Learning and Natural Language Processing. He's highly proficient with Python and NodeJS.