False positives caught in spam filter

  • 1
  • Problem
  • Updated 1 year ago
  • Not a Problem
In the past week, we've gotten multiple applications caught in the spam filter at 97-99% confidence who were definitely real students - had previous applications on file with us or had taken classes here in the past. One went all the way back to 2016. 

I've been monitoring these for about a year and there have only been a handful who were not from first time applicants - can the spam filter be updated to not catch students who have already submitted an app from the same CCCID? 
Photo of Lora Blakley

Lora Blakley

  • 18 Posts
  • 0 Reply Likes

Posted 1 year ago

  • 1
Photo of Joel Diaz

Joel Diaz

  • 32 Posts
  • 5 Reply Likes

I have encountered the same thing with a couple of applications.
Photo of Merrie Wales

Merrie Wales, Official Rep

  • 960 Posts
  • 57 Reply Likes
Joel,
Have you used the utility to mark the valid applications as "Valid"?  The SPAM filter has no way to know/understand that an application (or student) has previously attended courses at any given College.  We do not have access to your SIS to be able to validate that a student has taken courses at your College.  

We have many "students", a single CCCID, that apply to multiple Colleges (and they may or may not be SPAM applications) per day, so having "applied" in the past isn't a good indicator of Valid or not Valid applications.  

You indicate that you have gotten multiple applications recently that were caught in the SPAM filter - how big a percentage of your applications last week were marked a potential SPAM that were Valid?  If you would like you can create a Zendesk ticket that contains the confirmation numbers for those "valid" applications and we can do some investigation.  It sounds like the SPAM filter is doing exactly what it is intended to do... notify you when there are potential SPAM applications and the A&R staff can verify whether they are Valid or not
Photo of Joel Diaz

Joel Diaz

  • 32 Posts
  • 5 Reply Likes
It has only been a few applications for OC but I have marked them as valid in the SPAM filter. The percentage is minimal at this point so no need to create a ticket for OC. If the number increases significantly, I will submit a ticket.

Thank You.
Photo of Merrie Wales

Merrie Wales, Official Rep

  • 960 Posts
  • 57 Reply Likes
Good morning Lora,
Have you used the utility to mark the valid applications as "Valid"?  The SPAM filter has no way to know/understand that an application (or student) has previously attended courses at any given College.  We do not have access to your SIS to be able to validate that a student has taken courses at your College. 

We have many "students", a single CCCID, that apply to multiple Colleges (and they may or may not be SPAM applications) per day, so having "applied" in the past isn't a good indicator of Valid or not Valid applications. 

You indicate that you have gotten multiple applications recently that were caught in the SPAM filter - how big a percentage of your applications last week were marked a potential SPAM that were Valid?  If you would like you can create a Zendesk ticket that contains the confirmation numbers for those "valid" applications and we can do some investigation.  It sounds like the SPAM filter is doing exactly what it is intended to do... notify you when there are potential SPAM applications and the A&R staff can verify whether they are Valid or not.
Photo of Lora Blakley

Lora Blakley

  • 18 Posts
  • 0 Reply Likes
I did flag them as valid on on the admin site. In the past week there have only been about 10 apps caught in the spam filter, 4 or 5 of which were valid students. My concern is that they were flagged as 97-99% confidence and I do not know why they have been flagged, especially since I've contacted the tech center before and been told to confirm them as spam if they have such a high confidence rating. 
Photo of Merrie Wales

Merrie Wales, Official Rep

  • 960 Posts
  • 57 Reply Likes
Lora, 
Perfect.  There are about 200 different things the filter looks at, and I cannot say specifically what made the matrix be 97-99 percent "SPAM" for your 4-5 valid application but the Machine Learning tool learns from each application submitted and identifies patterns in submissions.  There will probably never be a tool that is 100% accurate.
Photo of Bill Abiko

Bill Abiko

  • 34 Posts
  • 2 Reply Likes
We too have had false positives and now check out each one carefully.  Most of the time they are obviously spam but not always.  Last month we sent email to a batch of applicants caught.  After no response for 10 days we marked them.  Then the responses came.

I know it's been asked but it would be nice to know why an applicant has been marked as spam with 97%+ certainty.  It often eludes us.  And I know this would be a much bigger project than is probably possible but it would be nice if we could set some criteria for possible spam before it gets downloaded.  One of the markers I look at is in a supplemental field which I imagine the filter can't consider.