This semester, two students at BYU-Idaho are working with FamilySearch to code and identify fraudulent information in data collected from Africa.

The FamilySearch fraud detection project helps FamilySearch to keep better track of the information collected by contractors and interviewers throughout Africa.

David Ouimette, a certified genealogist, said in his project syllabus that FamilySearch interviewed elders who know the oral genealogy of their people. These interviews started in 2003 and every year since, these efforts have increased and more oral genealogies were shared.

Ouimette also explains that the importance of this project in Africa stems from how little genealogy is written down is and instead recited orally to remind the people where they come from. Writing this information down helps preserve the genealogy and helps everyone know where they are from.

The problem with receiving all of this data comes when contractors and interviewers cheat the system in order to receive more money from FamilySearch.

“So with the FamilySearch project, our end goal for this semester is to identify what submissions are good and what submissions have fake names in them,” said Mike Min, a senior studying mathematical sciences at BYU-I.

Min has worked on this project for two semesters. The project works with data collected by interviewers over the past year.

Last semester, Min focused on finding corrupt contractors who turned in fraudulent information. This helped FamilySearch fire those contractors and help others be more honest in their work.

Mike Min describing the FamilySearch fraud detection project. Photo credit: Rose Jones.
Mike Min describing the FamilySearch fraud detection project. Photo credit: Rose Jones.

At the end of last semester, there were three tests created to determine the validity of the information gathered.

Min and LaRena Allen, a senior studying mathematics and science, have worked 20 hours a week this semester to find a way to identify false interviews among the many collected from Africa.

“Most of the semester, I’ve been working on a measure of how different an interview is from the country’s average interview,” Allen said. “My algorithm has been quantifying how weird an interview is and also how weird an interviewer is compared to other interviewers.”

These students are taking the time necessary to find out which interviewers are cheating the worst and which ones FamilySearch can still work with, trusting that they will be more honest in their interviews.

This project, to find detection, takes a lot of work and dedication from Min and Allen.

Mike Min and LaRena Allen working on their computers. Photo credit: Rose Jones.
Mike Min and LaRena Allen working on their computers. Photo credit: Rose Jones.

“When the information is given, there is hidden information waiting to be extracted and waiting to be unfolded,” Min said. “That’s a painful process, and sometimes it’s long and tiring, but once you actually help unfold some of the hidden information, things make more sense.”

The algorithms that these students have found cannot be disclosed because FamilySearch does not want contractors to find ways to stay out of detection when they submit family information. 

“I will say that there are multiple variables that greatly help us identify which contractor is cheating,” Min said. “And it will require some data wrangling and programming.”