Beyond the Coverage of Information Spreading: Analytical and Empirical Evidence of Re-exposure in Large-scale Online Social Networks

Xin Lü, Shuo Qin, Petter Holme, Fanhui Meng, Yanqing Hu, Fredrik Liljeros, Gad AllonarXiv 2019 · DOI · arXiv

Abstract

Peer influence and social contagion are key denominators in the adoption and participation of information spreading, such as news propagation, word-of-mouth or viral marketing. In this study, we argue that it is biased to only focus on the scale and coverage of information spreading, and propose that the level of influence reinforcement, quantified by the re-exposure rate, i.e., the rate of individuals who are repeatedly exposed to the same information, should be considered together to measure the effectiveness of spreading. We show that local network structural characteristics significantly affects the probability of being exposed or re-exposed to the same information. After analyzing trending news on the super large-scale online network of Sina Weibo (China's Twitter) with 430 million connected users, we find a class of users with extremely low exposure rate, even they are following tens of thousands of others; and the re-exposure rate is substantially higher for news with more transmission waves and stronger secondary forwarding. While exposure and re-exposure rate typically grow together with the scale of spreading, we find exceptional cases where it is possible to achieve a high exposure rate while maintaining low re-exposure rate, or vice versa.

Introduction: The Reach vs. Reinforcement Dilemma

In the world of viral marketing and news propagation, "reach" (or exposure rate) is traditionally considered the primary metric of success. However, this study argues that reach is a biased metric if considered in isolation. For information to truly "stick" and drive behavioral change—such as joining a movement or purchasing a product—social contagion theory suggests that "social reinforcement" is often necessary. This reinforcement occurs when an individual is repeatedly exposed to the same information from multiple sources.

Conceptual framework of exposure and re-exposure in social networks, illustrating how users receive redundant information from multiple publishers.

The authors address the lack of a formal, large-scale analytical framework linking network topology to this "information redundancy." By defining the re-exposure rate ($R^{re}$), they seek to understand why some news items flood our screens while others barely make a ripple, and how network structure can be leveraged to strategically control this redundancy.

Methodology and Datasets

The researchers employed a two-pronged approach. First, they developed analytical models using Random Walk and Branching Process theories to derive mathematical approximations for the probability of a node being exposed ($R^{exp}$) or re-exposed ($R^{re}$) based on its degree and surrounding network structure. Second, they validated these models against four real-world social networks:

  • Facebook: A small-scale network of 4,039 nodes.
  • Deezer: A European music streaming friendship network (41,773 nodes).
  • Wiki-vote: A network of Wikipedia users voting for adminship (7,115 nodes).
  • Sina Weibo: The centerpiece of the study, featuring 430 million active users and 51.8 billion links, covering 50 trending topics from 2017.

Empirical Insights from Sina Weibo

Analyzing the massive Sina Weibo dataset revealed several critical phenomena regarding how information penetrates large-scale networks:

The "Bot" Dip: A surprising "sudden drop" in exposure rates was found for users following between 2,048 ($2^{11}$) and 8,192 ($2^{13}$) people. These accounts are often automated or commercial bots. Despite their high out-degree, they are "socially distant" and isolated from the organic clusters where information circulates, leading to near-zero re-exposure.

Type I vs. Type II News: The paper categorizes news into two distinct spreading patterns. Type I (e.g., celebrity posts) starts with a massive burst from a central influencer but dies out quickly with low re-exposure. Type II (organic viral news) spreads more slowly but penetrates deeper into the network, achieving re-exposure rates 20-30% higher than Type I.

Empirical analysis of Sina Weibo data, highlighting the exposure rate dip for specific out-degrees (C, D) and the differences in propagation depth and re-exposure between Type I and Type II news (E, F, G).

Key Results and Strategic Implications

The study demonstrates that while node degree is positively correlated with re-exposure, local clustering is the dominant factor in determining if a user sees a post twice. High network clustering acts as a barrier to long-distance spreading but serves as the primary engine for social reinforcement.

Furthermore, the authors proposed a "Common Neighbor" (CN) strategy to decouple exposure from re-exposure. By choosing to spread information through links that share many common neighbors (CN-max), they can maximize re-exposure for a given reach. Conversely, using links with few common neighbors (CN-min) allows information to spread more thinly but broadly, maximizing coverage while minimizing redundancy.

Conclusion

This research shifts the focus from simple coverage to the quality of spreading through re-exposure. While marketers may want to maximize re-exposure for reinforcement, platforms may seek to minimize it to prevent "information overload." Understanding these structural dynamics allows for the strategic design of information campaigns that can either penetrate deep into social clusters or sweep broadly across the network landscape.

Summary updated

Abstract

Peer influence and social contagion are key denominators in the adoption and participation of information spreading, such as news propagation, word-of-mouth or viral marketing. In this study, we argue that it is biased to only focus on the scale and coverage of information spreading, and propose that the level of influence reinforcement, quantified by the re-exposure rate, i.e., the rate of individuals who are repeatedly exposed to the same information, should be considered together to measure the effectiveness of spreading. We show that local network structural characteristics significantly affects the probability of being exposed or re-exposed to the same information. After analyzing trending news on the super large-scale online network of Sina Weibo (China's Twitter) with 430 million connected users, we find a class of users with extremely low exposure rate, even they are following tens of thousands of others; and the re-exposure rate is substantially higher for news with more transmission waves and stronger secondary forwarding. While exposure and re-exposure rate typically grow together with the scale of spreading, we find exceptional cases where it is possible to achieve a high exposure rate while maintaining low re-exposure rate, or vice versa.
BibTeX
@misc{lu2019beyond,
  author = {Xin Lü and Shuo Qin and Petter Holme and Fanhui Meng and Yanqing Hu and Fredrik Liljeros and Gad Allon},
  title = {{Beyond the Coverage of Information Spreading: Analytical and Empirical Evidence of Re-exposure in Large-scale Online Social Networks}},
  year = {2019},
  eprint = {1907.12389},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1907.12389},
}
Figure