Introducing Causality and Traceability in Word-of-Mouth Algorithms
msra
摘要
Understanding the spread of information in a social network has proven useful in numerous areas, with viral marketing and epidimeology being two of the more prominent ones. Word-of- mouth algorithms are one class of algorithms that have been developed to model how information is verbally spread in a social network. However, two significant limitations of current word-of- mouth algorithms are their inability to: (1) capture when communication or contacts take place and (2) explain where the information possessed by each individual came from. In this paper, we present a novel word-of-mouth algorithm that addresses these drawbacks by considering the temporality of communication and by tracing the spread of influence within a social network. The traces of influence prove useful for the identification of the most important individuals in a social network and for inferring causality. We apply the proposed algorithm to a large set of call detail records (CDRs) and validate it via simulations of word-of- mouth traces. Our two main findings are that (1) influence is better understood when the temporal dimension is added to the model and (2) the spread of information and influence in a network has several statistical invariants.
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络