[Paper Review] The power of indirect social ties

Xiang Zuo, Jeremy Blackburn+3 more|arXiv (Cornell University)|Jan 17, 2014

Social Capital and Networks34 references3 citations

TL;DR

This paper proposes a social strength metric to quantify indirect social ties—relationships between individuals connected through one or more mutual friends—by combining interaction frequency and multiplicity of connecting paths. The metric improves friend-to-friend storage recruitment and predicts information diffusion with up to 93.5% accuracy, 2–3 steps in advance, enhancing resource availability and enabling proactive intervention in social networks.

ABSTRACT

While direct social ties have been intensely studied in the context of computer-mediated social networks, indirect ties (e.g., friends of friends) have seen little attention. Yet in real life, we often rely on friends of our friends for recommendations (of good doctors, good schools, or good babysitters), for introduction to a new job opportunity, and for many other occasional needs. In this work we attempt to 1) quantify the strength of indirect social ties, 2) validate it, and 3) empirically demonstrate its usefulness for distributed applications on two examples. We quantify social strength of indirect ties using a(ny) measure of the strength of the direct ties that connect two people and the intuition provided by the sociology literature. We validate the proposed metric experimentally by comparing correlations with other direct social tie evaluators. We show via data-driven experiments that the proposed metric for social strength can be used successfully for social applications. Specifically, we show that it alleviates known problems in friend-to-friend storage systems by addressing two previously documented shortcomings: reduced set of storage candidates and data availability correlations. We also show that it can be used for predicting the effects of a social diffusion with an accuracy of up to 93.5%.

Motivation & Objective

• To quantify the strength of indirect social ties, which are often overlooked despite their real-world utility in recommendations, job referrals, and information sharing.
• To validate the proposed social strength metric against established tie strength measures like Jaccard index and interaction frequency.
• To demonstrate the metric’s utility in two practical applications: enhancing friend-to-friend storage systems and predicting social diffusion paths.
• To address known limitations in friend-to-friend systems, such as limited candidate pools and low data availability due to synchronization issues.
• To enable decentralized, scalable, and privacy-preserving social network applications through localized computation of social strength.

Proposed method

• The social strength metric computes indirect tie strength as a function of the number of 2-hop paths and the strength of direct ties along each path, using interaction frequency as a proxy for tie strength.
• It models asymmetry in social relationships, reflecting that A may perceive B as a strong indirect tie even if B does not reciprocate the perception.
• The metric incorporates multiplicity of paths: a user connected via multiple 2-hop routes is considered more socially proximate than one with only a single path.
• It uses real-world datasets (CA-I, CA-II, TF2) to compute and validate social strength against correlation with Jaccard index and interaction frequency.
• For prediction, the metric ranks potential diffusion paths based on social strength, enabling early prediction of infection or information spread up to 3 steps ahead.
• Baseline comparisons use uniform acceptance assumptions across all 2- and 3-hop neighbors, contrasting with the metric’s adaptive, strength-based prediction.

Experimental results

Research questions

• RQ1 How can the strength of indirect social ties be quantitatively measured in a way that reflects real-world social dynamics and asymmetry?
• RQ2 To what extent does the proposed social strength metric correlate with established measures of direct tie strength, such as Jaccard index and interaction frequency?
• RQ3 Can the social strength metric improve the recruitment of storage candidates in friend-to-friend backup systems by increasing available resources and reducing dependency on synchronized availability?
• RQ4 How accurately can social strength predict the path of information diffusion in social networks, especially 2–3 steps before infection occurs?
• RQ5 In what ways can social strength enhance decentralized, privacy-preserving applications in social networks?

Key findings

• The social strength metric shows strong positive correlation with both the Jaccard index and interaction frequency, validating its accuracy in measuring indirect tie strength.
• Using social strength for storage recruitment increased online availability of resources by up to 20% by expanding the candidate pool beyond direct friends.
• The metric achieved a prediction accuracy of up to 93.5% in forecasting information diffusion paths, with sensitivity, specificity, and accuracy all exceeding 0.74 in all tested networks.
• Even for 3-hop predictions, the metric maintained performance above 0.5 in sensitivity, specificity, and accuracy, outperforming a uniform acceptance baseline across all three networks (CA-I, CA-II, TF2).
• The social strength-based prediction method significantly outperformed the baseline, especially in partially diffused networks, with peak performance at 0.89 sensitivity, 0.79 specificity, and 0.82 accuracy.
• The metric’s effectiveness was consistent across networks of varying density, indicating robustness to structural differences.