Versions Compared

Old Version 15

changes.mady.by.user Patricia Donohue

Saved on

compared with

New Version 16

changes.mady.by.user Patricia Donohue

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

The purpose of the Spam Filter Web Service is to assist colleges in making accurate and informed decisions on whether an application coming in through CCCApply is fraudulent or not. At the heart of the web service is a machine learning, continuous-training model that does NOT make decisions, it just predicts whether an application meets specific "identifiers" that have been collected and analyzed by the model based on thousands of confirmed spam applications submitted by the colleges. 

...


Note

New Retraining Model Released - May 10.  The retraining phase of the spam filter machine learning system was enabled in production on Friday, May 10. Retraining allows the machine learning system to learn and adapt from the changes in fraud signatures. What's Next? In the next phase, the machine learning team will focus on integration of PII features into the model. This will further improve the performance of the model in its ability to detect fraudulent applications. This iteration will be released at the end of June 2019.

The User Interface tool gives the colleges the ability to review each application flagged as fraud and then make the final decision on whether they should be confirmed as fraud or not. The continuous learning and retraining of the model is based on the final confirmation by the college.

...