A Personalized Assistant Framework for Service Recommendation

The popularity of service-oriented computing makes more and more services available on the Web. Users make use of these services to achieve their personal goals, such as purchasing movie tickets on-line and booking flights. Existing research has proposed various techniques to assist users to select services for achieving user goals. Typically, user choice of services change under different contexts. However, these approaches cannot recommend the desired services based on the changes of user contexts, and are not able to learn from user service selection history. In this paper, we provide an intellectually cognitive personalized assistant framework to achieve user goals. In particular, considering user contexts and historical service selection, our framework interacts with users by asking relevant and necessary questions, and help users navigate through sets of services to identify the desired services. We have designed and developed a prototype as a proof of concept. We perform a case study to evaluate the effectiveness of our framework. On average, our framework, utilizing the learning-to-rank algorithm, namely AdaRank, improves the nine baseline approaches by 12.02%-31.52% in helping users find the desired services. Our user study results show that our framework is helpful in achieving user goals and useful in saving users' time in finding their personalized services faster.