Media capacity estimation: Discovering news feed accounts for information sourcing

Micro blogging is gaining increasingly popularity for both publishing updates and receiving news. However, the enormous daily generated statuses are far beyond human´s reading limitation. “Follow” mechanism partly reduces the information overloaded problem but introduces a new one: it is hard to select whom to subscribe from vast accounts. Our objective is to find the accounts that are suitable to be subscribed as news feeds. We propose a media capacity model that measures accounts´ ability using criteria as: activity, reliability, content amount and posting stability. Besides, Weighed Borda Counting model is used to make collective decision. Furthermore, a supervised weight learning approach is designed to assign the weights for each criterion. The proposed method is evaluated on a one-year sampling of Sina postings. Experimental results show that the approach is able to top out accounts that gain better synthetic performance than others under selected criteria, and could obtain an average precision of 93.58% for top 200 users.