Split the cash from cache-friendly recommendations

Abstract

Recommender systems have been established as a key component of video streaming services, shaping up to 80% of content requests. Hence, recommendations are employed by the Content Providers (CPs) of these services to increase the viewing time and their revenues. Furthermore, it has been recently suggested that recommendations could be a means to reduce the operational costs of the Content Delivery Networks (CDNs) when they are related to already cached items, i.e., when they are cache-friendly. Clearly, these conflicting objectives, i.e., increasing revenue for the CP and reducing costs for the CDN, can create tensions between the two entities, and hence, prevent the full utilization of recommendations. In this work, we propose a model for capturing these tradeoffs, and an economic mechanism, based on the Nash bargaining solution, for reconciling the potentially conflicting objectives of the CP and the CDN. Our scheme enables the CP and CDN to jointly design the recommendations in a way that balances the revenue gains and cost savings, ensuring a fair and Pareto optimal split of the accrued benefits for both entities. Our numerical experiments in realistic scenarios show that the proposed scheme leads to important financial gains of up to 30%.• We propose a novel approach for using recommendations in OTT services using a hybrid revenue -cost criterion.1 Another interesting direction would be to model a collaboration where the two entities decide on both the recommendations and caching allocation. However, such an approach would require a better (and non-trivial) coordination among the two entities that we plan to address in future work.

Quid pro Quo in Streaming Services: Algorithms for Cooperative Recommendations

“…This paper extends our earlier work [1] by providing indepth insights on the proposed cooperation scheme, extending the cooperation mechanism to a distributed algorithm and presenting a comprehensive evaluation of the proposed algorithms in a variety of scenarios and for different input parameters. Finally, the current work discusses a possible extension of the presented problem towards the CDN's caching decisions.…”

“…101017109 (DAEMON). Part of this work appeared in the proceedings of the IEEE Global Communications Conference (GLOBECOM) 2021 [1].…”

“…• follows the recommendations with probability α u ; where, w.l.o.g. each of the N u items is considered equally likely to be requested 1 . Hence, each recommended content is requested by the user with probability α u /N u .…”

“…Therefore, the baseline recommendations Y b are the ones with the highest relevances r ui per user. Finally, it is worth noting that, in [1], our numerical evaluations were performed for R ui being a concave function of r ui that could similarly capture an adbased, subscription-based, or hybrid revenue model. Under this assumption, we obtained similar performance results as the ones presented here.…”