San Diego State Slashes Its Transcript Evaluation Time
Shutterstock.com
San Diego State University has reduced its
transcript evaluation process from eight weeks to two days, even during peak
periods, by implementing a new scanning process that automatically reads and
records transcript data faster and more accurately than before.
Optical Character Frustration
When the university receives transcripts from potential students, it needs to
enter the information from those transcripts into its student information and degree audit systems, to determine which, if any, of
the student's previous coursework can be applied to their degree at San Diego
State. In the past, the university used a combination of manual data entry and
optical character recognition software called OCR for Anydoc to get the data into their SIS and DAS. However, both the manual
and OCR processes were time-consuming and
error-prone.
The key to using OCR to automatically scan data from a transcript into an
information system is to create templates to map the individual data points on a
transcript to specific fields in the information system. For example, the
student's name on the transcript should map to the student's name in the SIS or
DAS, as should other information about the student, courses and grades on the
transcript.
According to Sue Reyes, programmer and analyst for enrollment services at San
Diego State, the process of scanning transcripts using OCR for Anydoc was
cryptic, and it was difficult to make new templates because the user interface
was not very graphical or intuitive. "It took quite a while to build a template,
get it set up correctly and get the data pulled off," she said. Consequently,
the university only went to the effort to build templates for schools from which
they frequently received transcripts, and they used a manual data entry process
for the rest.
OCR for Anydoc was integrated with the university's imaging system, Hershey
Singularity. When staffers scanned a transcript using OCR for Anydoc, it would also
send an image of the transcript to Hershey Singularity to keep on file for
future reference. But a few years ago, another company, called
Hyland Software, bought Hershey Singularity. Faced
with ongoing frustration with OCR for Anydoc, which was supported through
Hershey, Reyes and her team had reached a crossroad.
"Hershey had the contract with OCR for Anydoc and all of the support and
integration went through Hershey's support," said Reyes. "Hershey Singularity
had been bought out, so we could stay with what we had, or we could take a look
around and see if we can get something in here that's going to be more
effective, more efficient."
Scan Plan
Reyes and her team wanted something with faster and easier template-making
capabilities, and with a higher accuracy rate. "OCR is an automated process, but
it's not a hands-free automated process," said Reyes. "It still requires a lot
of manual intervention. You still need someone to look at it and verify it, but
the goal was to do that more quickly and effectively to cut down the number of
corrections we had to make."
They also needed an OCR system that would let them capture special flags,
such as whether a student had repeated a particular course and whether it should
be included in the GPA. "All of those little flags are sent in a million
different ways by the colleges," said Reyes. "There's no standard format to send
those notations. So in our OCR, we really wanted to capture not only the course
data and the grades, but also those special notations up front in the OCR
process."
Reyes and her team spent a couple of months evaluating different
systems to find one that would let them make templates quickly and easily to
find the variations in transcript codes, grab those values and put them into the
correct fields in the output file for the degree audit system. Throughout this
process they worked with Hyland, and discovered that the company's OnBase system
could handle those anomalies in data.
Rapid Transcript
San Diego State implemented Hyland Advanced Capture over a three-week period.
During the implementation process, Hyland staff came to the campus
to train the university's transcript processing staff on site for a few days, and then
the university started setting up templates. Reyes wrote about five Visual Basic
scripts that attached to the transcripts to let them capture inconsistent
values. Now, when a staff member needs to create a new template, they just
attach one of the scripts to it, and it automatically picks up the values
they're looking for and flags the courses appropriately.
"When I got the staff who had been using the older tools and had been doing
the manual data entry to learn how to build the templates, they were impressed,"
said Reyes. "It was something that was very intimidating in the old system, and
they realized how quickly and easily it could be done in the new system."
The university now has more than 1,100 transcript templates set up in the new
system. When the university receives a transcript, staffers check to see if they
have a scanning template for that school. If not, they create a new one, a
process that now requires only 10 minutes at most. Once the transcript is
scanned, the data verifiers open it in the OnBase indexing module to ensure the
process transferred the data correctly. The transcript is then sent to the OnBase
imaging system, so there's an image of the transcript on file, and the data from
the OCR process is imported to the student information and degree audit
systems. The degree audit system tries to match the courses from the transcript
to San Diego State courses to see if the credits will transfer to the degree the
student is applying for at the university. The next morning, the student can log
on to San Diego State's Web portal to see his or her transcript and which
courses did or did not transfer to the university.
"They know right off the bat, a day or two after they turn in their
transcript," said Reyes.
Even in the spring, when the university is receiving hundreds of transcripts
each day, it only takes a couple of days to scan and verify them, whereas before
it would take the university eight weeks or longer to process them.
"It's never going to be a completely automated process, but by leveraging the
scripts, the regular expressions, the power of the template, you can really cut
a lot of time out of processing of transcripts and increase the accuracy," said
Reyes. "I think that's for us the biggest thing — increasing that accuracy.
Even if it took the same amount of time, the fact that it's so much more
accurate going through the OCR process is a tremendous boost."