Deep learning based web service recommendation methods: A survey

What is it about?

The Web services recommender systems field have not yet benefited from all the advantages of Deep Learning techniques, compared to recommender systems in other fields (movies, music, products, …etc.). However, recommender systems and Deep Learning provide many advantages in the discovery, selection, composition and substitution of Web services. Properly exploiting these last operations, allows developers to derive maximum benefit from the reuse of Web services and their collaboration in the form of composite services called Mashups. Therefore, a good state-of-the-art analysis of DL-based Web services recommendation is necessary to help researchers, especially since the Web services area has its specificities and many differences with other recommendation systems domains. From our point of view, the three existing surveys onWeb services recommendation are too restricted and don’t include works based on DL.

Featured Image

Photo by Mohamed Nohassi on Unsplash

Photo by Mohamed Nohassi on Unsplash

Why is it important?

In this context, we present a comprehensive survey and analysis of studies on DL-based Web service recommender systems. For this, we have adopted the SLR (Systematic Literature Review) method. And the research questions (RQ) that our review will answer are the following: RQ1: What are the problems faced by recommender systems in the Web services field? RQ2: What are the Deep-Learning techniques used to build service recommendation systems, and to solve these problems? RQ3: What are the metrics used to evaluate the performance of Web services recommendation systems based on Deep Learning? RQ4: What are the future research directions for Web service recommender systems based on Deep Learning? To answer these questions, our survey provides the following steps: - Study several works on Web service recommender systems based on Deep Learning, using SLR method. - Present main challenges faced by recommender systems for Web services, then the solutions brought to recommender systems to face these challenges using Deep Learning techniques. - Classify studied works according to several criteria, namely: recommendation strategies, Web services properties taken into account, DL techniques used, data sets and performance evaluation metrics. - Present quantitative analysis of these works and give guidelines that will help researchers in this field by considering research shortcomings. - Provide a reliable background for our future contributions in this area.

Perspectives