ThinkMind // International Journal On Advances in Intelligent Systems, volume 3, numbers 3 and 4, 2010 // View article intsys_v3_n34_2010_13
A Framework for Progressive Trusting Services
Keywords: recommenders; reputation; dynamic feedback; services similarity; temporal similarity; use profiles, dynamic patterns
The web-based transactions, web services, and service-oriented platforms require appropriate mechanisms to announce, select, and use different services. A user is always under the dilemma of ‘use and trust’ or ‘trust and use’ for different services based on the notion of service reputation. The interaction between every service provider and its users is regulated by the service level agreement and customer satisfaction feedback. The former is the basis for the technical audit, while the latter subjectively validates the user perception. Selection of a most appropriate service by correctly invoking is a challenge. This is due to the difficulties to correctly expose proper way to invoke a service, to the variety of services, from on-line services, software pieces, to shopping, and to different invoker behavior. When considering invoker feedback, service ranking based on user perception, or based on recommenders’ statistics are relevant. A significant aspect is played by service similarity. The paper presents a framework and appropriate mechanisms to evaluate the services/providers in the light of their respective direct impact on user perception. To accurately evaluate the feedback after service/product consumption, we will refine the user profile by considering the dynamics of the feedback. The approach we propose deals with peaks in feedbacks. We consider quick negative and quick positive feedback as well as late vs. early feedback with respect to the time of the transaction. We propose formal concepts used in selecting an appropriate service. The paper presents adapted approaches to select services based on distance and similarity, and introduces a similarity taxonomy to better tune various kinds of service invocation under specific constraints, such feature relaxation, type of similarity, context, and service ranking. Selection is based on the feedback from the user. The proposed model is used for building a selection algorithm that allows variations on service invocation. The proposal is validated through use cases.
Pages: 326 to 346
Copyright: Copyright (c) to authors, 2010. Used with permission.
Publication date: April 6, 2011
Published in: journal