UIST: Predicting Tie Strength With Social Media

Article: Predicting Tie Strength With Social Media

Authors: Eric Gilbert and Karrie Karahalios

Summary: This article delves into the attempts by the scientific community to be able to evaluate "tie strength" (How close one user is to another, how much they trust someone) via certain variables that arise from interaction between the two users. With the rise in popularity of social networking, like LinkedIn or Facebook, the value of this research and its potential result, a predictive model for tie strength, has dramatically risen in recent years. The authors first describe the background of the "tie strength" concept, which originated in the 70s, and has only recently become more of a hot issue. The authors wrote the article in an attempt to gather some quantitative data about "tie strength" and perform a proof-of-concept, showing that the theory of "tie strength" is a relevant one to both individuals and businesses. In order to gather this data, they first identified a target social networking site, and they chose Facebook. They then combed through Facebook in order to identify certain variables that could be "potential indicators" of "tie strength". They found 74 of these variables, each within different categories of the "tie strength" concept, such as "Wall words exchanged", "Mutual friends", "Inbox message length", etc. They gathered data from these variables as well as from a questionnaire that was posed to 35 participants that were asked to participate in the research. The participants were given 30 minutes to answer the questions about as many facebook friends as possible. The authors then ran this data through their algorithm, and came up with results, both suprising and not suprising, about the predicted "tie strength" given by the algorithm, and the actual "tie strength" according to the participants. The study was able to come up with an 85% accuracy at determining "weak ties" (acquaintances, business associates) and "strong ties" (family, close friends).

Shaun's Opinion: The research done in this field by these two authors is pretty interesting. While I don't see much value in this research for me personally, I can see how it would potentially be useful to politicians, who want to get a message out to people of similar background, experiences, and viewpoints. I think the most interesting part of this whole study is the fact that the authors were able to identify 74 potential "tie strength" indicators, just in facebook alone. That is quite a staggering number to identify, I think if the task were given to me I could've come up with 10 or so.