In recent decades the amount of data and variety of information has grown rapidly. data storage, compression, and analyzing these data has become a vital subject in data mining and Machine Learning. the accuracy of the compression without losing the important data in this process needs attention. Therefore, this work presents an effective data compression for recommender systems based on the attention mechanism. In this method, the data compression is performed on two levels, features, and records. The technique is time aware and based on time windows and gives weight to users’ activity and will prevent losing important data. The result of this technique can be efficiently utilized for deep networks in which the amount of data is one of the serious problems. The experimental results show that the help of this technique not only reduces the amount of data and process time but also can reach acceptable accuracy.

Keywords