Document worth reading: “Deep Learning-based Sequential Recommender Systems: Concepts, Algorithms, and Evaluations”

In the sphere of sequential suggestion, deep learning methods have acquired quite a lot of consideration so far few years and surpassed typical fashions equal to Markov chain-based and factorization-based ones. However, DL-based methods even have some essential drawbacks, equal to insufficient modeling of shopper illustration and ignoring to distinguish the assorted sorts of interactions (i.e., shopper conduct) amongst clients and devices. In this view, this survey focuses on DL-based sequential recommender strategies by taking the aforementioned factors into consideration. Specifically, we illustrate the concept of sequential suggestion, counsel a categorization of present algorithms by the use of three types of behavioral sequence, summarize the necessary factor components affecting the effectivity of DL-based fashions, and conduct corresponding evaluations to show the results of these components. We conclude this survey by systematically outlining future directions and challenges on this self-discipline. Deep Learning-based Sequential Recommender Systems: Concepts, Algorithms, and Evaluations