Predicting tie strength with social media

CHI, pp. 211-220, 2009.

Cited by: 1645|Bibtex|Views38|Links
EI
Keywords:
social media design elementpredictive modeltie strengthtopic social sciencesocial mediaMore(12+)
Weibo:
To roughly adjust for it, all of the results presented here cut the degrees of freedom in half, a technique borrowed from the social networks literature

Abstract:

Social media treats all users the same: trusted friend or total stranger, with little or nothing in between. In reality, relationships fall everywhere along this spectrum, a topic social science has investigated for decades under the theme of tie strength. Our work bridges this gap between theory and practice. In this paper, we present a ...More

Code:

Data:

0
Introduction
  • Social science has made much the same case, documenting how different types of relationships impact individuals and organizations [16].
  • In this line of research, relationships are measured in the currency of tie strength [17].
  • Known as weak ties, can help a friend generate creative ideas [4] or find a job [18].
  • A link between actors either exists or not, with the relationship having few properties of its own [1, 2, 27]
Highlights
  • Using theory to guide the selection of predictive variables, we present the construction of our tie strength model
  • To roughly adjust for it, all of the results presented here cut the degrees of freedom in half, a technique borrowed from the social networks literature [33]
  • The Structural dimension plays a minor role as a linear factor
  • One way to read this result is that individual relationships matter, but they get filtered through a friend’s clique before impacting tie strength
  • Our results show that social media can predict tie strength
Methods
  • The authors modeled tie strength as a linear combination of the predictive variables, plus terms for dimension interactions and network structure: More complex models were explored, but a linear model allows them to take advantage of the full dataset and explain the results once it is built.
  • Si represents the tie strength of the ith friend.
  • To the best of the knowledge, exploring the interactions between the dimensions of tie strength is a novel approach
Results
  • Because each participant rated more than one friend, observations within a participant were not independent.
  • On average the model predicts tie strength within one-tenth of its true value.
  • This error interval tightens near the ends of the continuum because predictions are capped between 0 and 1.
  • The Structural dimension plays a minor role as a linear factor
  • It has an important modulating role via these interactions.
  • One way to read this result is that individual relationships matter, but they get filtered through a friend’s clique before impacting tie strength
Conclusion
  • The authors' results show that social media can predict tie strength. The How strong? model predicts tie strength within onetenth of its true value on a continuous 0–1 scale, a resolution probably acceptable for most applications.
  • The Intimacy dimension makes the greatest contribution to tie strength, accounting for 32.8% of the model’s predictive capacity.
  • This parallels Marsden’s finding that emotional closeness best reflects tie strength [33].
  • As Duration illustrates, a single variable can account for a large part of the model’s predictive capacity.In this paper, the authors have revealed a specific mechanism by which tie strength manifests itself in social media.
  • How do users relate to one another in these spaces? Do the data left behind tell a consistent story, a story from which the authors can infer something meaningful? The authors think this work takes a significant step toward definitively answering these questions
Summary
  • Introduction:

    Social science has made much the same case, documenting how different types of relationships impact individuals and organizations [16].
  • In this line of research, relationships are measured in the currency of tie strength [17].
  • Known as weak ties, can help a friend generate creative ideas [4] or find a job [18].
  • A link between actors either exists or not, with the relationship having few properties of its own [1, 2, 27]
  • Objectives:

    This paper aims to bridge the gap, merging the theory behind tie strength with the data behind social media.
  • The authors' goal was to collect data about the friendships that could act, in some combination, as a predictor for tie strength
  • Methods:

    The authors modeled tie strength as a linear combination of the predictive variables, plus terms for dimension interactions and network structure: More complex models were explored, but a linear model allows them to take advantage of the full dataset and explain the results once it is built.
  • Si represents the tie strength of the ith friend.
  • To the best of the knowledge, exploring the interactions between the dimensions of tie strength is a novel approach
  • Results:

    Because each participant rated more than one friend, observations within a participant were not independent.
  • On average the model predicts tie strength within one-tenth of its true value.
  • This error interval tightens near the ends of the continuum because predictions are capped between 0 and 1.
  • The Structural dimension plays a minor role as a linear factor
  • It has an important modulating role via these interactions.
  • One way to read this result is that individual relationships matter, but they get filtered through a friend’s clique before impacting tie strength
  • Conclusion:

    The authors' results show that social media can predict tie strength. The How strong? model predicts tie strength within onetenth of its true value on a continuous 0–1 scale, a resolution probably acceptable for most applications.
  • The Intimacy dimension makes the greatest contribution to tie strength, accounting for 32.8% of the model’s predictive capacity.
  • This parallels Marsden’s finding that emotional closeness best reflects tie strength [33].
  • As Duration illustrates, a single variable can account for a large part of the model’s predictive capacity.In this paper, the authors have revealed a specific mechanism by which tie strength manifests itself in social media.
  • How do users relate to one another in these spaces? Do the data left behind tell a consistent story, a story from which the authors can infer something meaningful? The authors think this work takes a significant step toward definitively answering these questions
Tables
  • Table1: Thirty-two of over seventy variables used to predict tie strength, collected for each of the 2,184 friendships in our dataset. The distributions accompanying each variable begin at zero and end at the adjacent maximum. Most variables are not normally distributed. The Predictive Variables subsection expands on some of these variables and presents those not included in this table
  • Table2: The five questions used to assess tie strength, accompanied by their distributions. The distributions present participant responses mapped onto a continuous 0–1 scale. Our model predicts these responses as a function of the variables presented in Table 1
  • Table3: The fifteen predictive variables with highest standardized beta coefficients. The two Days since variables have large coefficients because of the difference between never communicating and communicating once. The utility distribution of the predictive variables forms a power-law distribution: with only these fifteen variables, the model has over half of the information it needs to predict tie strength
  • Table4: The intercorrelations of the five dependent variables. With the exception of Job-Strong, Job-Loan and Bring-Job, the dependent variables are well-correlated with one another
Download tables as Excel
Funding
  • This work was funded in part by NSF Grant 0643502 and by Microsoft
Reference
  • Adamic, L. A. and Adar, E. 2003. Friends and neighbors on the Web. Social Networks, 25(3), 211–230.
    Google ScholarLocate open access versionFindings
  • Albert, R. and Barabási, A. L. 2002. Statistical Mechanics of Complex Networks. Reviews of Modern Physics, 74(1).
    Google ScholarLocate open access versionFindings
  • Bernard, H. R., Killworth, P., et al. 1984. The Problem of Informant Accuracy: The Validity of Retrospective Data. Annual Review of Anthropology, 13, 495–517.
    Google ScholarLocate open access versionFindings
  • Burt, R. S. 2004. Structural Holes and Good Ideas. American Journal of Sociology, 110(2), 349–399.
    Google ScholarLocate open access versionFindings
  • Burt, R. Structural Holes: The Social Structure of Competition. Harvard University Press, 1995.
    Google ScholarFindings
  • Chopra, S., Thampy, T., et al. Discovering the Hidden Structure of House Prices with a Non-parametric Latent Manifold Model. Proc. KDD, 2007. 173–182.
    Google ScholarLocate open access versionFindings
  • Clark, H. H. Arenas of Language Use. University Of Chicago Press, 1993.
    Google ScholarFindings
  • Constant, D., Sproull, L., et al. 1996. The Kindness of Strangers: The Usefulness of Electronic Weak Ties for Technical Advice. Organization Science, 7(2), 119–135.
    Google ScholarLocate open access versionFindings
  • Cunningham, W. H., Cunningham, I. C., et al. 1977. The Ipsative Process to Reduce Response Set Bias. Public Opinion Quaterly, 41(3), 379–384.
    Google ScholarLocate open access versionFindings
  • Frakes, W. B. and Baeza-Yates, R. Information Retrieval: Data Structures and Algorithms. Prentice Hall, 1992.
    Google ScholarFindings
  • Friedkin, N. E. 1980. A Test of Structural Features of Granovetter's Strength of Weak Ties Theory. Social Networks, 2, 411–422.
    Google ScholarLocate open access versionFindings
  • Gergle, D., Kraut, R. E., et al. The Impact of Delayed Visual Feedback on Collaborative Performance. Proc. CHI, 2006. 1303–1312.
    Google ScholarLocate open access versionFindings
  • Gilbert, E., Karahalios, K., et al. The Network in the Garden: An Empirical Analysis of Social Media in Rural Life. Proc. CHI, 2008. 1603–1612.
    Google ScholarLocate open access versionFindings
  • Golder, S., Wilkinson, D. M., et al. Rhythms of Social Interaction: Messaging Within a Massive Online Network. Proc. 3rd International Conference on Communities and Technologies (CT2007), 2007.
    Google ScholarLocate open access versionFindings
  • Google Scholar. http://scholar.google.com/scholar?q=weak+ties. Accessed September 16, 2008.
    Findings
  • Granovetter, M. 1983. The Strength of Weak Ties: A Network Theory Revisited. Sociological Theory, 1, 201–233.
    Google ScholarLocate open access versionFindings
  • Granovetter, M. S. 1973. The Strength of Weak Ties. The American Journal of Sociology, 78(6), 1360–1380.
    Google ScholarLocate open access versionFindings
  • Granovetter, M. Getting a Job: A Study of Contacts and Careers. University Of Chicago Press, 1974.
    Google ScholarFindings
  • Greasemonkey. http://www.greasespot.net. Accessed September 16, 2008.
    Findings
  • Hansen, M. T. 1999. The Search-Transfer Problem: The Role of Weak Ties in Sharing Knowledge Across Organization Subunits. Administrative Science Quarterly, 44(1), 82–111.
    Google ScholarLocate open access versionFindings
  • Haythornthwaite, C. 2002. Strong, Weak, and Latent Ties and the Impact of New Media.. Information Society, 18(5), 385–401.
    Google ScholarLocate open access versionFindings
  • Haythornthwaite, C. and Wellman, B. 1998. Work, Friendship, and Media Use for Information Exchange in a Networked Organization. J. Am. Soc. Inf. Sci., 49(12), 1101–1114.
    Google ScholarLocate open access versionFindings
  • Krackhardt, D. The Strength of Strong Ties: The Importance of Philos in Organizations. In N. Nohria and R. Eccles (Ed.), Networks and Organizations: Structure, Form and Action (216–239). Boston, MA: Harvard Business School Press.
    Google ScholarLocate open access versionFindings
  • Krackhardt, D. and Stern, R. N. 1988. Informal Networks and Organizational Crises: An Experimental Simulation. Social Psychology Quarterly, 51(2), 123– 140.
    Google ScholarLocate open access versionFindings
  • Lampe, C., Ellison, N., et al. Changes in Participation and Perception of Facebook. Proc. CSCW, 2008. 721– 730.
    Google ScholarLocate open access versionFindings
  • Laumann, E. O., Gagnon, J. H., et al. 1989. Monitoring the AIDS Epidemic in the United States: A Network Approach. Science, 244(4909), 1186–1189.
    Google ScholarLocate open access versionFindings
  • Liben-Nowell, D. and Kleinberg, J. The Link Prediction Problem for Social Networks. Proc. of the 12th international conference on Information and knowledge management, 2003. 556–559.
    Google ScholarLocate open access versionFindings
  • Lin, N., Dayton, P. W., et al. 1978. Analyzing the Instrumental Use of Relations in the Context of Social Structure. Sociological Methods Research, 7(2), 149– 166.
    Google ScholarLocate open access versionFindings
  • Lin, N., Ensel, W. M., et al. 1981. Social Resources and Strength of Ties: Structural Factors in Occupational Status Attainment. American Sociological Review, 46(4), 393–405.
    Google ScholarLocate open access versionFindings
  • Marin, A. 1981. Are Respondents More Likely To List Alters with Certain Characteristics?: Implications for Name Generator Data. Social Networks, 26(4), 289– 307.
    Google ScholarLocate open access versionFindings
  • Marsden, P. V. 1981. Core Discussion Networks of Americans. American Sociological Review, 52(1), 122– 131.
    Google ScholarLocate open access versionFindings
  • Marsden, P. V. 1990. Network Data and Measurement. Annual Review of Sociology, 16(1), 435–463.
    Google ScholarLocate open access versionFindings
  • Marsden, P. V. and Campbell, K. E. 1990. Measuring Tie Strength. Social Forces, 63(2), 482–501.
    Google ScholarLocate open access versionFindings
  • Pennebaker, J. W. and Francis, M. E. Linguistic Inquiry and Word Count. Lawrence Erlbaum, 1999.
    Google ScholarFindings
  • Rapleaf. RapLeaf Study of Social Network Users vs. Age. http://business.rapleaf.com/company_press_2008_06_18.html. Accessed September 16, 2008.
    Findings
  • Schaefer, C., Coyne, J. C., et al. 1990. The Health-related Functions of Social Support. Journal of Behavioral Medicine, 4(4), 381–406.
    Google ScholarLocate open access versionFindings
  • Shi, X., Adamic, L. A., et al. 2007. Networks of Strong Ties. Physica A: Statistical Mechanics and its Applications, 378(1), 33–47.
    Google ScholarLocate open access versionFindings
  • Tahmincioglu, E. Facebook Friends As Job References? http://www.msnbc.msn.com/id/26223330. Accessed September 16, 2008.
    Findings
  • Uzzi, B. 1999. Embeddedness in the Making of Financial Capital: How Social Relations and Networks Benefit Firms Seeking Financing. American Sociological Review, 64(4), 481–505.
    Google ScholarLocate open access versionFindings
  • Wellman, B. and Wortley, S. 1990. Different Strokes from Different Folks: Community Ties and Social Support. The American Journal of Sociology, 96(3), 558–588.
    Google ScholarLocate open access versionFindings
  • Whittaker, S., Terveen, L., et al. The Dynamics of Mass Interaction. Proc. CSCW, 1998. 257–264.
    Google ScholarLocate open access versionFindings
Your rating :
0

 

Best Paper
Best Paper of CHI, 2009
Tags
Comments