Over the past two years, the CCCApply team has been working to combat the high volume of suspicious applications submitted by malicious actors using a wide variety of evolving techniques to obtain .edu email accounts. Some colleges receive up to 10,000 suspicious applications a day.

Working with a team of data scientists, the CCCApply team implemented the Spam Filter Web Service in September 2018, which uses a machine learning, continuous-improvement retraining model to predict fraud and suspend suspicious applications before they reach the colleges. When spam is suspected, the application is sent to a CCCApply Administrator holding folder where college staff can review and mark fraudulent applications accordingly. As colleges provide feedback, the spam filter learns to identify certain markers to more accurately discern submission validity.

A May 2019 spam filter update improved CCCApply’s ability to detect actual fraud and reduce the number of false positives. “The machine learning team is now focused on integrating personally identifiable information (PII) features into the model to further improve detection of fraudulent applications. This iteration will be released at the end of June 2019,” said Patricia Donohue, CCCApply Product Manager. 

Automatic .edu Email Creation Not Recommended

While the spam filter continues to reduce the number of fraudulent applications getting through, it is recommended that colleges discontinue automatic creation of a .edu address upon submission. Instead, colleges should activate an account only after the student takes another step in the onboarding process such as completing orientation, meeting with a counselor or registering for a class.

“Colleges that have taken steps to remove these practices have seen a sharp decline in incoming fraud; in some cases, fraud stopped altogether,” said Tim Calhoon, Executive Director of the CCC Technology Center.

More information about the Spam Filter Web Service is available in the CCCApply Public Documentation resources.

 

Similar Posts