Building FCA-Based Decision Trees for the Selection of Heterogeneous Services

Abstract

Industry Track 4 - Services Selection and CompositionInternational audienceLate-binding and substitutability offered by the service-oriented approach improve adaptability but increase the need for fast and efficient algorithms to select services. In this paper, we proposed to use the Formal Concept Analysis (FCA) approach as a classification tool to select services at runtime, according to user specifications. We propose to classify existing services and generate a decision tree to help user select the most appropriate service(s). One of advantages of using FCA is the ability to select without additional cost an equivalent service in the case of a service must be replaced at runtime. Our approach have been implemented and validated on pervasive use cases within a European collaborative project

fANFARE: Autonomic Framework for Service-Based Pervasive Environment

“…In [9], [10], we propose to apply FCA to service selection in pervasive environment. The usability of the FCA is based on the computation of the interesting concepts of the concept lattice.…”

Practical Use of Formal Concept Analysis in Service-Oriented Computing

“…Services are classified according to their characteristics. In [7], we have extended our approach to the composition of decision structures in response to complex user requests. …”

“…Consequently, many technologies have been implemented and adapted to these domains. For instance, UPnP 7 or DPWS 8 are preferred in small area networks for devices whereas OSGi 9 and iPOJO [8] are often used in centralized and embedded equipments. Service-Oriented Computing has thus evolved to support dynamic service discovery and lazy inter-service binding.…”