Factors Dominating Individuals' Retweeting Decisions

摘要

Individuals' retweeting action is the driving force of information dissemination on social networking sites, without which tweets cannot arrive at audiences other than the source node's followers. Existing research tends to predict individuals' retweeting decisions with more and more factors/features, without examining the relevance of them and even why these factors are added is rarely discussed. As the number of factors grows up, the existence of irrelevant/redundant factors causes significant problems. Undoubtedly, finding dominating factors can deepen our understanding about individuals' retweeting decisions and save the cost of measuring redundant factors and computing resources. In this paper, we focus on exploring the relative importance of these factors with multiple statistical and data mining algorithms, and finally identify dominating factors which substantially affect individuals' retweeting decisions. Experiments demonstrate the superior performance of adopting only dominating factors in contrast with adopting full features set.