Information Diffusion and External Influence in Networks (1206.1331v1)

Abstract: Social networks play a fundamental role in the diffusion of information. However, there are two different ways of how information reaches a person in a network. Information reaches us through connections in our social networks, as well as through the influence of external out-of-network sources, like the mainstream media. While most present models of information adoption in networks assume information only passes from a node to node via the edges of the underlying network, the recent availability of massive online social media data allows us to study this process in more detail. We present a model in which information can reach a node via the links of the social network or through the influence of external sources. We then develop an efficient model parameter fitting technique and apply the model to the emergence of URL mentions in the Twitter network. Using a complete one month trace of Twitter we study how information reaches the nodes of the network. We quantify the external influences over time and describe how these influences affect the information adoption. We discover that the information tends to "jump" across the network, which can only be explained as an effect of an unobservable external influence on the network. We find that only about 71% of the information volume in Twitter can be attributed to network diffusion, and the remaining 29% is due to external events and factors outside the network.

Information Diffusion and External Influence in Networks

This paper addresses the dynamics of information diffusion within social networks, with a specific focus on distinguishing between information transmission via network connections and external sources. The authors develop a model that incorporates both internal diffusion mechanisms and external influences, such as mass media, to paper how information spreads, particularly within the Twitter network.

Model and Methodology

The authors present a probabilistic generative model that allows information to reach nodes through two channels: internal network links and external influences. The model acknowledges that traditional analyses often overlook the impact of external sources. This is critical, as approximately 29% of Twitter information volume is found to be influenced by external events, challenging the predominant node-to-node diffusion assumption.

Key components of the model include:

• Internal Hazard Function: Governs the time interval for information to pass from an infected node to its neighbors.
• Event Profile $\lambda_{ext}(t)$: Represents the time-varying probability of external exposures, capturing the presence of unobservable external influences.
• Exposure Curve $\eta(x)$: Translates the number of exposures into the likelihood of infection, thereby mapping exposure dynamics and probability of dissemination across the network.

The inference process involves iteratively estimating the exposure curve and event profile using a combination of predefined infection times and network structure, moving towards an optimal parameterization through convergence.

Empirical Analysis

Extensive experiments with synthetic and Twitter data validate the model. Synthetic data experiments demonstrate the model's robustness in accurately inferring parameters compared to simpler baseline methods.

In the real data analysis, the researchers apply the model to Twitter, specifically analyzing the spread of URLs. The inferred event profiles align well with known external events, demonstrating the model's capability to detect exogenous influences effectively. For instance, the model accurately identifies spikes in event profiles corresponding to specific external triggers in the Tucson, Arizona shooting case paper.

Key Findings and Implications

Several insights emerge from the analysis:

1. Extent of External Influence: On average, 29% of URL exposures on Twitter are attributed to external sources. This finding underscores the significance of considering non-network influences in information diffusion models.
2. Topic-Specific Insights: The model reveals differences in external influence across categories. Politics and Sports are notably influenced by external sources, whereas Technology and Entertainment are more internally driven.
3. Exposure Dynamics: The paper provides empirical evidence suggesting a high selectivity in idea adoption, as inferred from the consistently low $\rho_1$ values, the peak probability of infection.

The proposed model not only distinguishes internal and external influences but also enhances the understanding of how information propagates within and outside the network structure. This has broader implications for developing strategies in network-based marketing, information control, and policy-making regarding the spread of information within digital environments.

Future Directions

The findings pose interesting questions for future exploration. Consideration of how individual node behaviors contribute to broader diffusion patterns could yield deeper insights into information dynamics. Furthermore, extending the model to address the heterogeneous nature of external influences across nodes could refine predictions and improve applications in identifying network influencers.

Overall, this paper contributes a nuanced understanding of the mechanisms driving information diffusion, revealing the critical role of external influences in network-based communication models.