Content recommendations for increased relevance in digital media

Important results from the project

Triggerbee has, with Vinnova´s co-financing, carried out a project to kick start Triggerbees AI journey. The aim of the project has been to develop skills and gain practical experience of AI and ML, and to develop a pilot software for content recommendations on websites. The project has turned out well and led to positive effects, knowledge and a foundation for our continued work within AI / ML.

Expected long term effects

Data collection and processing were carried out and experiments of applying machine learning models to our data. An important insight is that our different customers have different data structures and this places special requirements on how the models are set up and trained, which makes it challenging to create as a generic commercial software, but that standard models and applications are very suitable. An important part has been to create these standard models and boundaries. This is also valuable for our upcoming recruitment of new developers to continue our AI work in the software.

Approach and implementation

** Denna text är maskinöversatt ** We created a solution that made it possible to create product recommendations on a predictive model, based on historical transaction data and information about read articles. The following scenarios are supported by our model: Article to article relationship: Type: “Readers who read this article also read often this article ". This increases the exposure of the article directory and the reading time on the site. The algorithm we chose is called SAR, (Smart Adaptive Recommendations).