دانلود رایگان مقاله انگلیسی تصمیم گیری مشکوک مبتنی بر شواهد در خرید آنلاین - اشپرینگر 2018

Abstract

Nowadays, Online shopping has been tremendous lifestyle choices due to the lower management cost for product/service providers and the cheaper prices for buyers/customers. Meanwhile, it raises a big challenge for both buyers and sellers to identify the right product items from the numerous choices and the right customers from a large number of different buyers. This motivates the study of recommendation system which computes recommendation scores for product items and filters out those with low scores. Recently, a promising direction involves the consideration of the social network influence in recommendation system. While significant performance improvement has been observed, it is still unclear to which extension the social network influence can help differentiate product items in terms of recommendation scores. This is an interesting problem in particular in the situation that the recommended product items have the highly similar (or identical) scores. As the first effort to this problem, this paper probes the boundary of social network influence to recommendation outputs by solving an optimization problem called evidence-driven dubious decision making. Two solutions have been proposed and the evaluation on two real world datasets has verified the effectiveness of the proposed solutions.

7 Conclusion

This work sheds light to understand the impact of social network influence in the field recommendation. We have proposed and addressed a significant problem, evidence-driven dubious decision making. It probes the boundary that social network influence can contribute to recommendation. With the flourish of social networks and the recommendation system in e-commerce, this boundary is essential since it is the benchmark to evaluate the method which exploring social network influence in recommendation. In this study, we point out this problem is NP-hard and two solutions have been developed. Their performance have been verified on two real world datasets.