Contextual Fuzzy Ranking for Web Services Discovery in a Hybrid Architecture

Introduction

Web services technology is a solution to software interoperability. The W3C (World Wide Web Consortium) defines a service as a software application, identified by a URI (Unified Resource Identifier), whose interfaces and connections are defined, described and discovered by XML artifacts (Roxin, 2014). The W3C specifies that interactions between services are direct and based on XML-type messages sent via Internet protocols.

Figure 1.

The Web services standards

The presented Web service (WS) model, allows the interaction between three actors: the service consumer (client), the service provider and the service discovery system. This interaction is made possible thanks to three functions: the publishing function, the discovery function and the binding function, described as follow:

• -

  The publishing function allows providers to publish descriptions of their services in order to make them available for client, in the UDDI registry.

• -

  The discovery system allows the client to search, in the UDDI registry, for services according to his requirements.

• -

  The binding function: when the discovery system presents to the user the list of discovered services, the user invokes one of the discovered services presented by the discovery system. The client and the provider will use SOAP protocol to communicate.

The used discovery technologies vary from an aspect to another: from registry deployment (centralized or distributed) to functional discovery (using keywords or semantic search) and non-functional discovery (using the context and the Quality of Service information) to get more precise and accurate results. The service discovery process is a key element in the Service-oriented architecture (SOA), although the increasing number of services put the reliability of the used technology of the discovery process in question.

In this paper, the authors propose a hybrid architecture based on a multiagent system (MAS), including a discovery process based on a contextual fuzzy ranking approach. The proposed architecture aims to solve some of weaknesses in the discovery process like latency, overload failure and network congestion. It is based on a distributed network using a single access point (Gateway), offers a solution for managing the increasing number of services using multi-level domain service distribution and an agent platform that manages the different tasks as publishing and discovery.

Also, authors propose as a second contribution a discovery agent implementing a contextual fuzzy ranking approach. This approach helps to choose the service offering the best level of quality parameters between services that have the same functionalities. Authors use the fuzzy ranking approach for contextual discovery which shows a promising result.

The paper is organized as follow: section two present some of the related works with discussions, section three explains the concept of the proposed hybrid architecture with details on how it works. Next, section four, gives details on the improvement of the contextual fuzzy ranking approach implemented in the hybrid architecture. Finally, the authors conclude the paper with a conclusion and some perspectives for their future work.