»Ê¼Ò»ªÈË

Tim Chartier came into national prominence through his development of algorithms to predict the outcome of NCAA basketball tournament games. National and international media contacted the »Ê¼Ò»ªÈË professor of mathematics and computer science about his "March Mathness" system, and countless individuals adopted and adapted it in their quest to beat the one-in-9.2-quintillion odds of correctly picking the outcome of every tournament game.

While no one has ever generated a perfect bracket, adopters of the algorithm created brackets that beat up to 99 percent of the millions of others submitted.

But more importantly than just giving people an edge in office pools, Chartier and collaborators have also found a way to use their sports "bracketology" for development of a potentially lucrative system of ranking content in social media. The system recently received U.S. Patent 9311620 as "A System and Process for Ranking Content on Social Networks Such as Twitter."

In accordance with the college's intellectual property policy, the patent officially belongs to »Ê¼Ò»ªÈË, rather than its three co-inventors–Chartier, his former student Lake Trask '11 and collaborator Professor Amy Langville at the College of Charleston. It is the first full patent ever granted to the college.

New Ranking Systems

Trask, who is pursuing a doctorate in operations research at N.C. State University, got involved as a summer research assistant to Chartier in 2010. The three of them worked on development of sports rankings systems, and soon turned to apply their work to new social networks, with a focus on Twitter. Langville has developed a national reputation for her analysis of how algorithms can rank data-an interest she expressed in a 2006 book titled, Google's PageRank and Beyond, the Science of Search Engine Rankings.

The team worked steadily on the ranking system for Twitter in 2010 and came to believe it might have commercial value. Following college protocol, Chartier and Trask presented the idea to the college intellectual property committee, which gave approval in April 2011 to apply to the U.S. Government Patent Office for a one-year provisional patent.

That step protected their idea while they refined it further and began exploring commercial applications.

"The trick with patenting is that you can't share the work publicly before you get the provisional patent," Chartier said. "The college moved quickly on this work because Lake was set to present it at a national conference. Securing the provisional patent enabled him to make the presentation, and for our work to move forward."

During the provisional patent period, a few businesses expressed interest in commercializing the system, but the »Ê¼Ò»ªÈË team didn't have the marketing expertise to devote themselves to promoting it.

Alumnus Gets Involved

Fortunately, they attracted another ally with the skills they needed. Investment banker Paul Ward '81, a former mathematics major and current managing partner of Clarety Global Investments, was thoroughly experienced in exploring investment opportunities for his firm, and as an alumnus, his interest was piqued by Chartier's system.

"Ranking pages on the web is really about reputation and authority, and Google has made billions of dollars auctioning words that can improve a page's reputation," said Ward. "The approach Tim and Amy devised is a truly novel approach that could have huge implications."

Clarety Global was already considering uses for algorithms that would mine and analyze the tremendous amount of data generated by social networks with applications in music recommendations, expert networks and influencer marketing. Ward recognized that »Ê¼Ò»ªÈË was approaching the idea in new ways that deserved full patent protection.

Ranking content of web pages is nothing new. The most prominent system, developed by Google, is "PageRank," which counts the number and quality of links to a web page to determine an estimate of a website's "importance."

But Chartier's group found that PageRank does not apply accurately to the content of a social network like Twitter, which is based on decisions to follow or not follow other users, rather than a surfer visiting and following links on web pages.

Logically, they figured, the quickest way to spread information would be through someone of great importance on the network, and their system helps identify and measure an individual's importance. This approach to an "influencer marketing" system based on sports ranking systems was new and exciting.

"Right now, companies are succeeding or failing based on how they pay their influencers. They're making it up as they go along," Ward said.

Social Network Users Provide Data

The patent points out, "The sheer amount of people who use online social networks regularly has turned them into important tools that can be used by advertisers and businesses... The invention provides a system and process to determine what users may be important compared to others."

"Previously there was no way to identify Twitter 'influencers' who have the strongest voices on the network," Chartier said. "Those influencers could be very important in marketing, for instance, to spread the word about a product."

"If someone who has your trust, or whose authority you respect, recommends that you buy something you're much more likely to act than if a company suggests you buy their products," Ward explained. "Person-to-person recommendations generate more than twice the sales of paid advertising. Identifying 'who influences who' is exactly what the »Ê¼Ò»ªÈË patent aims to solve. It also creates a framework for rationalizing customer engagement–how much you have to pay to generate engagement, how much it's worth, how to do it better, and how to do it faster. All kinds of interesting and valuable things flow out of this work."

The »Ê¼Ò»ªÈË team's innovative means of identifying influencers, primarily developed by Trask, came from treating data within a social network as a network of athletic contests.

"We had already developed the way to use an algorithm to rank college football and basketball teams," he recalled. "Then we realized that if we could model interactions between social media users as games, we could come up with a system that could rank the influence of each user. If we could rank the influences within a group or network as a whole, we would have something advertisers would value as a way to identify top influencers and get them to talk about your product."

Simply viewed, a person who is followed by another person on Twitter "wins" the game. A person who chooses to follow another person "loses" the game. Users who accumulate the highest scores have the greatest influence on the network.

The inventors envision further development within aspects of Twitter. For instance, the system could order users by their importance, or recommend following users with similar interests. Also, outcomes of "games" could be affected not only by tweets, but also mentions, retweets, "strength of schedule," favorites and unfollows.

The patent also suggests that the system is applicable not only to Twitter, but also to other social media like Snapchat, Facebook and Instagram.

Ideas for Commercialization

Ward pointed out another idea for commercializing the system. He said a top five online streaming music company has been innovative in allowing listeners to submit their own playlists for others to enjoy. The system now features many highly successful DJs who create very popular playlists. But the company would benefit from software tools to identify and exploit the influence of the popular DJs.

"A technology like this could open up entirely new strategies for how they attract DJs, cultivate DJs, measure the impact of these DJs in the community and on the market, and–most important for any music service–how they can monetize their playlists," Ward observed.

Ward also believes that the system has value not only for commercial interests, but also a wide range of affinity groups.

"In the next 20 years in areas like social justice and the environment, collective action across political boundaries and across age groups will be increasingly necessary to effect change," he said. "These ad hoc social networks are all going to rise or fall based on how well they can engage their audiences. So if you can use the »Ê¼Ò»ªÈË system to create a resonant and sustainable mission-based social network, you might be able to make a difference much more rapidly and cost-effectively than if you just let it organize itself."