Seeding Network Influence in Biased Networks and the Benefits of Diversity

The problem of social influence maximization is widely applicable in designing viral campaigns, news dissemination, or medical aid. State-of-the-art algorithms often select “early adopters” that are most central in a network unfortunately mirroring or exacerbating historical biases and leaving under-represented communities out of the loop. Through a theoretical model of biased networks, we characterize the intricate relationship between diversity and efficiency, which sometimes may be at odds but may also reinforce each other. Most importantly, we find a mathematically proven analytical condition under which more equitable choices of early adopters lead simultaneously to fairer outcomes and larger outreach. Analysis of data on the DBLP network confirms that our condition is often met in real networks. We design and test a set of algorithms leveraging the network structure to optimize the diffusion of a message while avoiding to create disparate impact among participants based on their demographics, such as gender or race.

[1]  Yong-Yeol Ahn,et al.  Twitter's Glass Ceiling: The Effect of Perceived Gender on Online Visibility , 2016, ICWSM.

[2]  Wei Chen,et al.  Efficient influence maximization in social networks , 2009, KDD.

[3]  Albert,et al.  Emergence of scaling in random networks , 1999, Science.

[4]  Michael Luca,et al.  Racial Discrimination in the Sharing Economy: Evidence from a Field Experiment , 2016 .

[5]  Claire Mathieu,et al.  Homophily and the Glass Ceiling Effect in Social Networks , 2015, ITCS.

[6]  Lu Hong,et al.  Groups of diverse problem solvers can outperform groups of high-ability problem solvers. , 2004, Proceedings of the National Academy of Sciences of the United States of America.

[7]  Timnit Gebru,et al.  Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification , 2018, FAT.

[8]  Nicole Immorlica,et al.  The Importance of Communities for Learning to Influence , 2017, NIPS.

[9]  David García,et al.  Bias in Online Freelance Marketplaces: Evidence from TaskRabbit and Fiverr , 2017, CSCW.

[10]  Ana-Andreea Stoica,et al.  Algorithmic Glass Ceiling in Social Networks: The effects of social recommendations on network diversity , 2018, WWW.

[11]  Nisheeth K. Vishnoi,et al.  Ranking with Fairness Constraints , 2017, ICALP.

[12]  Ana-Andreea Stoica,et al.  Fairness in Social Influence Maximization , 2019, WWW.

[13]  Jon M. Kleinberg,et al.  Inherent Trade-Offs in the Fair Determination of Risk Scores , 2016, ITCS.

[14]  Krishna P. Gummadi,et al.  On the Fairness of Time-Critical Influence Maximization in Social Networks , 2019, IEEE Transactions on Knowledge and Data Engineering.

[15]  Markus Strohmaier,et al.  Homophily and minority size explain perception 1 biases in social networks , 2018 .

[16]  C. Hill,et al.  Why So Few? Women in Science, Technology, Engineering, and Mathematics. , 2010 .

[17]  Éva Tardos,et al.  Maximizing the Spread of Influence through a Social Network , 2015, Theory Comput..

[18]  Nathan Srebro,et al.  Equality of Opportunity in Supervised Learning , 2016, NIPS.

[19]  Carlos Eduardo Scheidegger,et al.  Certifying and Removing Disparate Impact , 2014, KDD.

[20]  Seth Ovadia,et al.  The Glass Ceiling Effect , 2001 .

[21]  Éva Tardos,et al.  Influential Nodes in a Diffusion Model for Social Networks , 2005, ICALP.

[22]  Masahiro Kimura,et al.  Tractable Models for Information Diffusion in Social Networks , 2006, PKDD.

[23]  Krishna P. Gummadi,et al.  Fairness Beyond Disparate Treatment & Disparate Impact: Learning Classification without Disparate Mistreatment , 2016, WWW.

[24]  S. Page.  The Difference: How the Power of Diversity Creates Better Groups, Firms, Schools, and Societies , 2007 .

[25]  M. Graham,et al.  Science faculty’s subtle gender biases favor male students , 2012, Proceedings of the National Academy of Sciences.

[26]  Eric Rice,et al.  Group-Fairness in Influence Maximization , 2019, IJCAI.

[27]  Pierre Fraigniaud,et al.  Preferential Attachment as a Unique Equilibrium , 2018, WWW.

[28]  Markus Strohmaier,et al.  Homophily and minority-group size explain perception biases in social networks , 2019, Nature Human Behaviour.

[29]  Wei Chen,et al.  Scalable influence maximization for prevalent viral marketing in large-scale social networks , 2010, KDD.

[30]  Sharad Goel,et al.  The Problem of Infra-Marginality in Outcome Tests for Discrimination , 2016, 1607.05376.

[31]  Sharad Goel,et al.  The Measure and Mismeasure of Fairness: A Critical Review of Fair Machine Learning , 2018, ArXiv.

[32]  Stanley Wasserman,et al.  Social Network Analysis: Methods and Applications , 1994, Structural analysis in the social sciences.

[33]  Lior Seeman,et al.  Adaptive Seeding in Social Networks , 2013, 2013 IEEE 54th Annual Symposium on Foundations of Computer Science.

[34]  Erez Shmueli,et al.  Algorithmic Fairness , 2020, ArXiv.

[35]  Jon M. Kleinberg,et al.  Selection Problems in the Presence of Implicit Bias , 2018, ITCS.

[36]  Wendy M. Williams,et al.  Why Aren?t More Women in Computer Science? , 2011, Making Software.

[37]  Danah Boyd,et al.  Gaps in Information Access in Social Networks? , 2019, WWW.

[38]  Eli Upfal,et al.  Stochastic models for the Web graph , 2000, Proceedings 41st Annual Symposium on Foundations of Computer Science.

[39]  Sune Lehmann,et al.  Understanding the Demographics of Twitter Users , 2011, ICWSM.

[40]  Sandra Wachter.  Affinity Profiling and Discrimination by Association in Online Behavioural Advertising , 2019, SSRN Electronic Journal.

[41]  Jie Tang,et al.  Influence Maximization in Dynamic Social Networks , 2013, 2013 IEEE 13th International Conference on Data Mining.

[42]  Michael Ley,et al.  DBLP - Some Lessons Learned , 2009, Proc. VLDB Endow..

[43]  Piotr Sapiezynski,et al.  Discrimination through Optimization , 2019, Proc. ACM Hum. Comput. Interact..