Predicting Attributes of User Profile in Social Networks by Analyzing Communities of their Ego-Network

In online social networks, a user is allowed to specify a lot of personal information - attributes. Some users provide only a part of whole information, or do not provide any information about themselves at all. Due to that, inferred hidden attributes are one of the fundamental problems of social analysis. The study proposes a new approach to user's hidden or unspecified attributes prediction. The method is based on analysis of the user's ego-network structure and attributes of its social graph vertices. The developed method was compared wth other methods according to three datasets of users' ego-networks from Facebook, Twitter and VKontakte social networks. It showed high values of F-measure, precision and completeness for predicting the chosen attributes of the user profile such as hometown or school. Using this method with additional data sources an analyst with high precision can reveal the identity of an anonymous social network user by their relations with other users.

References

[1] Davis C.A.J., Pappa G.L., de Oliveira D.R.R., Arcanjo F. de Zima. Inferring the location of twitter messages based on user relationships. T. GIS., 2011, vol. 15, no. 6, pp. 735-751. DOI: 10.1111/j.1467-9671.2011.01297.x Available at: http://onlinelibrary.wiley.com/doi/10.1111/j.1467-9671.2011.01297.x/abstract

[2] Li R., Wang C., Chang K. C.-C. User profiling in an ego network: co-profiling attributes and relationships. Proc. 23rd Int. Conf. on World Wide Web. WWW ’14. New York, ACM, 2014, pp. 819-830. Available at: http://wwwconference.org/proceedings/www2014/proceedings/p819.pdf

[3] Dong Y., Yang Y., Tang J., Yang Y., Chawla N.V. Inferring user demographics and social strategies in mobile social networks. Proc. 20th АСМ SIGKDD Int. Conf. on Knowledge Di5covery and Data Mining. KDD ’14. New York, ACM, 2014, pp. 15-24.

[4] Korshunov A., Beloborodov I., Buzun N., Avanesov V., et al. Social network analysis: methods and applications. Trudy Instituta sistemnogo programmirovaniya RAN [Proceedings of ISP RAS], 2014, vol. 26, no. 1, pp. 439-456 (in Russ.). Available at: http://cyberleninka.ru/article/n/analiz-sotsialnyh-setey-metody-i-prilozheniya

[5] Mislove A., Viswanath B., Gummadi K.P., Druschel P. You are who you know: inferring user profiles in online social networks. Proc. 3d ACM Int. Conf. on Web Search and Data Mining. WSDM ’10. New York, ACM, 2010, pp. 251-260.

[6] Chaabane A., Acs G., Kaafar M. You are what you like! Information leakage through users’ interests. Proc. Annual Network and Distributed System Security Symposium, 2012.

[7] Kosinski M., Stillwell D., Graepel T. Private traits and attributes are predictable from digital records of human behavior. Proc. of the National Academy of Sciences, 2013, vol. 110, no. 15, pp. 5802-5805. DOI: 10.1073/pnas.1218772110 Available at: http://www.pnas.org/content/110/15/5802.full

[8] Dougnon R.Y., Fournier-Viger P., Nkambou R. Advances in artificial intelligence. Proc. 28th Canadian Conf. on Artificial Intelligence. Canada, Springer International Publishing, 2015, pp. 84-99.

[9] Yang J., Leskovec J. Community-affiliation graph model for overlapping network community detection. 12th IEEE Int. Conf. on Data Mining, ICDM 2012. 2012, pp. 1170-1175. DOI: 10.1109/ICDM.2012.139 Available at: http://ieeexplore.ieee.org/document/6413734

[10] Chesnokov V.O. Intersecting communities allocation in social graphs based on majoritarian neighbours characteristic. LOMONOSOV-2016. XXIII Mezhd. nauch. konf. studentov, aspirantov i molodykh uchenykh [LOMONOSOV-2016. XXIII Int. conf. of students, postgraduates and young scientists]. Moscow, MAKS Press Publ., 2016, pp. 49-51.

[11] Clauset A., Newman M.E.J., Moore C. Finding community structure in very large networks. Phys. Rev. E., 2004, vol. 70, no. 6, pp. 1-6. DOI: 10.1103/PhysRevE.70.066111 Available at: http://journals.aps.org/pre/abstract/10.1103/PhysRevE.70.066111

[12] Rosvall M., Bergstrom C.T. Maps of random walks on complex networks reveal community structure. Proc. of the National Academy of Sciences, 2008, vol. 105, no. 4, pp. 1118-1123. DOI: 10.1073/pnas.0706851105 Available at: http://www.pnas.org/content/105/4/1118.full

[13] Yang J., Leskovec J. Overlapping community detection at scale: a nonnegative matrix factorization approach. Proc. of the 6th ACM Int. Conf. on Web Search and Data Mining. WSDM ’13. New York, 2013, pp. 587-596. DOI: 10.1145/2433396.2433471 Available at: http://dl.acm.org/citation.cfm?doid=2433396.2433471

[14] Yang J., McAuley J.J., Leskovec J. Community detection in networks with node attributes. 2013 IEEE 13th Int. Conf. on Data Mining. 2013, pp. 1151-1156.

[15] Leskovec J., Krevl A. SNAP datasets: Stanford large network dataset collection. Stanford Network Analysis Project: website. Available at: https://snap.stanford.edu/data (accessed 12.01.2017).

[16] Chesnokov V.O., Klyucharev P.G. Social graph community differentiated by node features with partly missing information. Nauka i obrazovanie: nauchnoe izdanie MGTU im. N.E. Baumana [Science and Education: Scientific Publication of BMSTU], 2015, no. 9, pp. 188-199 (in Russ.). DOI: 10.7463/0915.0811704