Accelerating DIN Model for Online CTR Prediction with Data Compression

As the key task of recommender systems, the click-through rate(CTR) prediction is to predict the probability of users clicking on a specific product. It is often costly due to the big data sets. In this paper, we apply some data compression technology to accelerate CTR prediction. By casting the data format to some more memory-efficient format, we can significantly improve a popular recommender method online.