Scanning the Social Media Minefield

UAH collaborates with a headhunter to create software that inspects candidates’ social media histories.

The Social Media 23 team, from left: Denise Medders, Andrea Hutchings, Alison Duncan and Alan Medders. Photo by Dennis Keim

Alan Medders is no stranger to the exhaustive vetting process that goes into hiring candidates for administrative and faculty positions. After all, he served 25 years as an administrator at public and private universities and colleges, as well as a stint working for Myers McRae, a firm based in Macon, Georgia, that performs searches for administrative and faculty positions for academic institutions across the country.

In early 2017, while working at Myers McRae, Medders had just completed a search for a university, and the candidate was introduced to the campus. “And, of course, when a new administrative staff or faculty member is announced, the first thing people do is Google the person,” says Medders. When that inevitably happened with the newly hired employee, a Facebook profile surfaced that falsely indicated he was a pedophile. The account was fake and quickly removed; fires were put out.

This particular situation was extreme, but not unfamiliar to Medders, who, during his tenure in academia, and increasingly in the last decade, had to terminate employees for content they posted on social media platforms that was viewed as problematic for their institutions. He had also witnessed multiple instances of people being denied employment because of their social media activities. As younger candidates with larger social media footprints enter the job market, manually vetting their profiles has become increasingly time-consuming. Medders wondered why there wasn’t an automated process, similar to standard employment background checks, already in place.

- Sponsor -

In June 2017, Medders teamed with his wife, Denise, to start their own academic headhunting firm, Higher Education Leadership Search, based in his hometown of Anniston. He contacted his daughter, Alison Duncan, a security analyst for the Alabama Law Enforcement Agency, and asked whether any software was available for his company to use for social media screening. There was, but nothing as comprehensive and accessible as Medders had in mind. He set out to see if a solution could be developed.

“I wanted to reach out to a research university and knew the quality of the computer science program at [the University of Alabama in Huntsville] and the work they’ve done with numerous companies in the Huntsville area and beyond,” says Medders.

He contacted Kannan Grant, director of the Office of Technology Commercialization at UAH, to see if the university had any interest in partnering with HELS. Grant, in turn, contacted computer science professors Harry Delugach and Haeyong Chung, and William “Ivey” MacKenzie, a human resources specialist from the College of Business. They met with Medders, who offered to provide funding for the research and formally agreed to move forward in creating a software program that would analyze the social DNA of a job candidate. Social Media 23 — alluding to 23 chromosomes — was born.

Over a period of about 18 months, the team of UAH faculty, along with two computer science graduate research assistants, worked on developing software that can automate the process of examining a potential employee’s profile on Facebook, Twitter and Instagram.

Duncan and Medder’s other daughter, Andrea Hutchings, who currently performs this service manually for HELS clients, says it typically takes them at least an hour per candidate to perform a social media background check. The Social Media 23 software reduces that process to one to two minutes.

The idea is that a candidate will consent to the vetting process, just as they would for a background check, allowing Social Media 23 temporary access to their accounts and profiles. “So they’re aware that we’re in there, looking at their pages, and we do it for a set period of time,” explains Medders.

The software then screens for five criteria: language, actions or affiliation with groups that discriminate based on race, religion or gender; obscene language or content in posts; use or abuse of controlled or illegal substances; and sexual harassment. More broadly, it also checks personal disposition, looking for language that expresses depression, anger or anxiety. The categories can be customized to suit a particular institution’s needs.

“Of course, all of the results will be examined to determine the nature of comments made in any of the five categories, whether obvious or requiring further analysis as to whether they are positive or negative when examined in context,” Medders explains. “Such as, ‘I am anxious about my test’ or ‘I am anxious about my interview.’  This is where algorithms in the software code and machine learning come into play that help provide an initial analysis.”

As the development took off, so did the UAH team’s enthusiasm for the software’s potential market.

“When we first started talking to them, we were simply talking about needing something for Higher Education Leadership Search,” says Medders. As the software progressed, the UAH team encouraged Medders to look beyond his own company. “They’re the ones who created the excitement and buzz well beyond what we were thinking about at the beginning of this process.”

Duncan and Hutchings, who have both been heavily involved in the development of Social Media 23, are also impressed with the level of enthusiasm and support they’ve received from the UAH team.

“You take two graduate students with their background in coding — and now they’re sending us information that we didn’t know about or didn’t understand how we could apply it — it’s inspiring. Having the conversations over dinner at our parents’ house to where we are today has been an amazing transformation,” says Duncan.

Medders says, “So it’s been an extremely positive experience, as I thought it would, because I understand higher ed, and I understand the research component of what they do.”

Kannan Grant agrees. “I think this is exactly one of the things UAH needs to be doing for people with ideas on the outside — help them develop it and give it back to them and say, ‘Go make a business out of this.’ This was exciting for UAH. The graduate students that worked on this project were exposed to real-world problems, and I think that’s a very valuable thing for our students, as well as our faculty.”

Social Media 23 entered the beta test phase in early September, with a handful of universities that have volunteered as subjects. The software platform is scheduled to roll out to clients in January 2019.

Katherine MacGilvray and Dennis Keim are freelance contributors to Business Alabama. Both are based in Huntsville.

The latest Alabama business news delivered to your inbox