In this blog post, we will demonstrate how we improved Pinterest Homefeed engagement volume from a machine learning model design perspective — by leveraging realtime user action features in Homefeed recommender system.
The Homepage of Pinterest is the one of most important surfaces for pinners to discover inspirational ideas and contributes to a large fraction of overall user engagement. The pins shown in the top positions on the Homefeed need to be personalized to create an engaging pinner experience. We retrieve a small fraction of the large volume of pins created on Pinterest as Homefeed candidate pins, according to user interest, followed boards, etc. To present the most relevant content to pinners, we then use a Homefeed ranking model (aka Pinnability model) to rank the retrieved candidates by accurately predicting their personalized relevance to given users. Therefore, the Homefeed ranking model plays an important role in improving pinner experience. Pinnability is a state-of-the-art neural network model that consumes pin signals, user signals, context signals, etc. and predicts user action given a pin. The high level architecture is shown in Figure 3.
To continue reading this article, click here.