A Context-Based Autonomous Construction Approach for Procedural Mashups

Wei He,Qingzhong Li,Lizhen Cui,Ting Li

DOI: https://doi.org/10.1109/icws.2014.75

2014-01-01

Abstract:Mashup is becoming a powerful approach for end-users to meet their ad-hoc requirements based on existing services. Quite a few researches have been performed to achieve rapid, on-demand, intuitive development of mashups, which mainly focus on finding suitable quality components from a large number of available services. However, for mashups with procedure and context features, it is more crucial and difficult to construct an effective mashup structure, rather than selecting individual components. In this paper, we propose a context-based autonomous construction approach for procedural mashup based on pattern mining. In our approach, the mashup composition process is divided into 2 phases: schema construction phase and component binding phase. First, context-based mashup schemas with probability are extracted and recovered by applying pattern mining tasks to historical mashup logs. Then, according to user goal and awareness of user context, an optimal mashup schema is composed progressively by top-k recommendations for the next behavior/activity, which will be grounded to Web-API based components later. The proposed approach can autonomously generate context-based mashup schema with quality components and high probability of success without dependence on user professionalism.