Who drive people to forward information: Publisher or spreader?

The explosive growth in online social networks makes them major platforms of information diffusion. People receive large-scale information from friends, but hardly find what they really want. Understanding influential factors of forwarding behavior can be used to improve the ranking algorithm, and prevent the information overload. In this paper, we compare the influence of the publisher and the spreader in Renren, the largest and oldest online social network in China. We crawl a connected graph component of 42.1 million users, 1.6 billion social relationships, and 118.2 million unique URLs. We compare URLs received from friends, and URLs which are really adopted. We observe that people are more influenced by spreaders than publishers: the spreader's recommendation time is more important than the publisher's publication time. People prefer URLs which were forwarded by spreaders a short time ago. Moreover, the previous adoption from the spreader is useful to predict user's forwarding behavior, while the previous adoption from the publisher is useless. These findings are useful to improve the ranking algorithm and prevent the information overload.