Optimized retrieval algorithms for personalized content aggregation

Personalized content aggregation methods, such as for news aggregation, are an emerging technology. The growth of mobile devices has only increased demand for timely updates on online information. To reduce traffic or bandwidth, efficient retrieval scheduling strategies have been developed to monitor new postings. Most of these methods, however, do not take user access patterns into consideration. For example, the strategy for a user who checks news once a day should be different from the strategy for a user who checks news ten times a day. In this paper, we propose a personalized content aggregation model in which delay time depends not only on the retrieval time and posting time, but also on user access patterns. With total expected delay as the objective, we derive a resource allocation strategy and retrieval scheduling strategy that is optimal when postings are Poisson. To our knowledge, this is the first personalized aggregation model on multiple data sources.