The Pluses & Minuses of Social Analytics


With the right algorithm at hand, you might be able to determine, not only who will Friend whom online, but also how one user likely feels about another. This is the gist of a study conducted by three computer science researchers, who predicted positive or negative relationships on three social sites by taking into account the relationships other users already have with each other.

For their study, the researchers examined Epinions, a review site that lets members express trust or distrust of other community members; Wikipedia, the user-edited online encyclopedia that lets members vote for or against another's nomination for site "adminship"; and Slashdot, a tech news site that lets participants label each other "Friend" or "Foe."

Specifically, the researchers -- Jure Leskovec of Stanford University and Daniel Huttenlocher and Jon Kleinberg of Cornell University -- relied on the social psychology theories of balance and status. Based on the principles "the enemy of my friend is my enemy," "the friend of my enemy is my enemy," and similar variations, they created an algorithm to calculate the unknown relationship between two individuals based on other relationships within the network.

Beyond simply predicting whether a particular user might attach a positive or negative link to another user, the researchers wanted to answer questions about why such decisions might be made within the network.

"Answers to these questions can help us reason about how negative relationships are used in online systems, and answers that generalize across multiple domains can help to illuminate some of the underlying principals," the researchers wrote in their paper, "Predicting Positive and Negative Links in Online Social Networks." They conclude that a "hidden" relationship between two members of a network can be inferred using an algorithm measuring the positive or negative attitudes of all others in the network.

Thus the research can truly be said to be a kind of sentiment analysis since it tries to gauge the sentiment of two individuals toward each other based on information about other sentiments in the network.

In a recent live chat debate between social media consultant Joe Stanganelli and Web analytics consultant Pierre DeBois, we looked at the question of whether sentiment analysis could provide the amount of data needed to be useful, given the level of accurate measurement currently possible. The researchers found that, using their algorithm at least, they could accurately predict positive and negative relationships between individuals. The results were fairly consistent over several different networks.

How could such an algorithm be used to predict sentiment for your company or organization, especially in online applications? Share your ideas on the boards below.

Shawn Hessinger, Community Editor

Shawn Hessinger is a community manager, blogger, social media and tech enthusiast, journalist, and entrepreneur based in Northeastern Pennsylvania. He serves as community manager and blogger for BizSugar.com, a business news and information Website, and contributes regularly to the online business news source, Small Business Trends. He is the founder of PostRanger.com, an online content and media community, and has provided blogging and social media services and consulting for companies all over the world. He researches and writes on a variety of business, Internet-related, and other tech topics including business intelligence and analytics. He is also keenly interested in computer-aided data management as it relates to his various online ventures. A newspaper journalist with more than 11 years experience as a reporter and then managing editor, Shawn began blogging in 2006 and now provides a variety of consulting and outsourcing services in Search Engine Optimization, Web development, and online marketing to companies large and small. He is a strong advocate for the use of BI and related computer data management in business decision making, whether using software as a service (SaaS), cloud, or other applications, and in the opportunity these technologies provide to transform small startups and larger established businesses alike.

BCBSNC, SAS Team on Advanced Analytics

The key to improving heathcare outcomes is to look at individual needs, the companies say.

Spoofing, Privacy Greatest Barriers for Biometrics

In Wednesday's e-chat, we discussed the analytics of identification and whether the technology might find a bigger role one day in marketing intelligence.


Match sites
  • 8/29/2011 10:05:06 AM
NO RATINGS

It seems like this could also be used in the context of online dating sites, to improve matches and the quality of match likelihoods.

Definitely the sort of thing the fellows at OKCupid would be interested in.

Tool for Law Enforcement
  • 8/28/2011 4:07:09 PM
NO RATINGS

Do you think this is something that can be used in law enforcement? Seems like this would be a way to "profile" or "watch" certain people based off this data. 

Re: Friends & enemies
  • 8/25/2011 11:17:04 AM
NO RATINGS

What would be really interesting to see is similar plus and minus vote features created for the larger sites like Facebook and Twitter. Such a change would definitely make a kind of sentiment analytics possible on a much larger scale because of the sheer size of these sites, but I fear it would also completely change their dynamic which is probably why it is unlikely ever to happen.

Re: Friends & enemies
  • 8/24/2011 7:46:16 AM
NO RATINGS

Ah, silly me, Shawn! -- mixing up my research here. So, yes, I do a greater connection to sentiment analysis on these sorts of sites with these sorts of choices than I do on "friending-oriented" sites like Facebook. Thanks for straightening me out!

Re: Friends & enemies
  • 8/23/2011 7:38:54 PM
NO RATINGS

Hi Beth,

In the case of the previous post on Facebook friendship choice you are definitely correct. However, in the current study, Slashdot members are being asked to label other members of the community "friends" or "foes", on Epinions, members are asked to decide whether they "trust" or "distrust" other members and on Wikipedia, members vote for or against other members' nominations for the position of "adminship." These definitely seem to be judgments that tell us something of the true sentiments at work in the network not simply arbitrary decision based on other factors.

Re: Friends & enemies
  • 8/23/2011 7:08:06 PM
NO RATINGS

Are we though, Shawn? The reasons I might not friend or accept friend requests may very well vary person by person. I might not like the person at all and want nothing to do with him or her. Or, I might have great professional respect and admiration for another person but really don't want him or her gaining glimpses of my personal life on Facebook. Or I might want to keep my Facebook circle to a select group, no matter how much I like others -- so, no cousins allowed or children's parents or some such. So while we're saying directly "we do or don't want to friend you" I don't think that's the same as saying, "we do or don't like you" or "we have a negative or positive sentiment about you." It's too gray in my mind.

Re: Friends & enemies
  • 8/23/2011 6:53:58 PM
NO RATINGS

Hi Beth,

Yes, this post has already been shared by the social analytics crowd via "Social Metrics and Analysis" and has pretty far ranging implications. You are right that this kind of "sentiment analytics" is far different from the approaches we were examining earlier. First, this is about positive and negative relationships between people, members of the online network specifically. But second, the results are far less open to interpretation. Users choose a positive or negative connection themselves thus we are not faced with a subjective interpretation of the semantics in a post but are told by the user directly what they think of other individuals in the network.

Friends & enemies
  • 8/23/2011 11:44:04 AM
NO RATINGS

Shawn, thanks for sharing another interesting look at what the university research crowd is doing relative to studying and predicting social media behavior. I think your take is interesting, too, that this ability to predict positive and negative relationships between individuals is a kind of sentiment analysis. Until now, I has always thought of the sentiments being analyzed as more overt in nature -- I'm so angry at X. Y is fantastic, etc.

 

 

INFORMATION RESOURCES
ANALYTICS IN ACTION
CARTERTOONS
VIEW ALL +
QUICK POLL
VIEW ALL +