Web Service Selection Based on QoS Prediction for Clustering and Ranking Services Using Auto-Encoder and K-Means

1. Introduction

There are many available Web services, and the number keeps on increasing. However, this ever- increasing pool makes it difficult for optimal service selection. Therefore, finding suitable Web services for users quickly and efficiently has become a challenging issue and a key problem to be solved (Hwang et al., 2014).

Several Web services may share similar functionalities but different non-functional properties. When discovering Web services, it is crucial to take into consideration functional and non-functional properties in order to render an effective and reliable service selection process (Rajendran et al., 2010). We can discover a Web service by its identifier on the Web and its input and output parameters, which are its functional information. However, other non-functional information, such as response time, availability, cost, reliability, and so on, can also be used to characterize the service. Thus, non-functional quality of service aspects is a crucial criterion for enhancing Web service selection (Ouadah et al., 2015).

Unfortunately, QoS differs from user to user, from context to context, and changes dynamically over time depending on various parameters (Ouadah et al., 2015). For example, the response time may vary depending on the amount of network traffic (Yu & Bouguettaya, 2010) and on the capacity of each service to maintain the same performance. Therefore, the challenge is to choose the best Web services from a set of functionally equivalent ones.

With the growing number of Web services, the complexity of selecting the best one will increase exponentially. So, selecting an appropriate Web service that suits user requirements is a non-trivial task, and recommender systems are developed to narrow down user choices. For effective recommendations, it is necessary to consider both functional and non-functional requirements. Functional requirements are about how a service works, while non-functional requirements are about QoS parameters. In recent years, the Web service selection based on the QoS requirements has received much attention in the service computing community. In this paper, we propose a Web service selection method for clustering and ranking services using auto-encoder and k-means. Our proposed method is based on our previous work for QoS prediction presented in (Merabet & Benmerzoug, 2021). The proposed method appears as a solution that helps users select a suitable Web service that meets their requirements.

In particular, our main contributions can be resumed as follows:

• •

  First, we use the K-Means algorithm to identify Web services with similar QoS values and group them into k clusters.

• •

  After the clustering phase, we rank the services in each cluster based on their average QoS values, from the best to the worst. Then, we recommend them to users.

• •

  We conducted a series of experiments with a real QoS dataset to validate and evaluate our method. The experimental results indicate that the proposed method is better performed. It can efficiently and accurately find the optimal service with optimal QoS values.

The remaining sections of this paper are as follows: Section 2 presents some related work. In section 3, we discuss our proposed framework. In sections 4 and 5, we show and discuss our results, respectively. Finally, section 6 concludes this paper and presents future directions.

2. Related Work

Several approaches are proposed in the context of service selection to solve the problem of changing service quality. However, most of these approaches are incompatible with critical processes due to the stringent global constraints they must respect. These tasks typically make decisions based on the QoS values of the selected services, but these values can be easily changed after a few moments of execution.

The problem of selecting Web services has recently received a lot of attention, and many researchers have used various algorithms to solve it. There are two types of Web service selection approaches, function-based and QoS-based (Dai, 2013).

Several approaches to service selection have been proposed in the literature. Furthermore, only a few noteworthy approaches are reviewed here.