Collaborative Approaches to Complementing Qualitative Preferences of Agents for Effective Service Selection

Abstract

Incomplete preferences of agents may render service selection ineffective. We address this problem by proposing a set of collaborative approaches to complementing agents' incomplete preferences in a qualitative way. For an agent, the approaches first find its similar agents, and then base the similar agents' qualitative preferences to complement this agent's missing preferences. We analyze and compare these approaches and provide experimental results to justify our arguments. We also compare our approach with the classic collaborative filtering and show the competitive advantages of our approach in service selection. Our work thus serves as an important step towards effective service selection.

Qualitative preference-based service selection for multiple agents

“…But in many real situations, a decision may have to be made for a group of agents with different preferences on QoS attributes of web services [4]. In addition, the agents' preferences may also be incomplete [5]. Incompleteness of preferences represents an absence of knowledge about the relationship between certain pairs of outcomes.…”