home > UCAS Implementation
UCAS requirement
CFL delivery
High volume, real-time detection. Turn round time of under 3 minutes for blocks of 125 submissions. 2500 - 5000 per hour depending on search base.
Compliance with UCAS' IT standards - Java™ and Oracle™>. Base program was written in Java™ and enhanced to meet all of UCAS' bespoke requirements, whilst enabling UCAS' in-house technical teams to fully integrate and support.
Handle large comparison datasets with a minimal deployment footprint. During 2008 operation over 1.5 million personal statements will be indexed and these indexes are used for all future comparisons - reducing the deployment footprint and overall hardware costs.
Full automation with normal operation leading to similarity detection within 10 minutes of an application being received at UCAS. The program runs on a timed basis, monitoring incoming data and processing it with no human intervention. To date the maximum wait time has been approximately 7 minutes.
Similarity transcripts that clearly identify areas within a personal statement that meet UCAS' detection criteria and display this information in a way that aids efficient and accurate decision making. Bespoke similarity transcripts developed that enable informed decisions to be made by UCAS staff and University admission officers.
Flexible detection criteria. UCAS can set the parameters relating to what and how much is indexed, what will be searched for and the amount of similarity required at sentence and document level to trigger a report. UCAS also have the ability to remove commonly used words from the trigger list - Duke Edinburgh Award etc.
Maximise UCAS' ROI on existing hardware and have proven reliability. The CopyCatch application is fully multi-threaded and exploits the capability of the Sun T2000 computer with its 8 cores, 32 threads. The main detection phase uses all 32 threads before collating the results for final comparison. During trials UCAS were able to keep their hardware running constantly for 9 days to prove the reliability of the application.
Seamless integration with an Oracle™ database. The program reads the Oracle™ database for incoming records and writes out any matches in full to a separate table, while updating the master record with the status of each personal statement.
Linguistic expertise throughout the development lifecycle. CFL were able to provide UCAS with guidance on linguistic issues relating to similarity detection during solution design, development and rollout. This ensured that UCAS got the right service for them without the need for their technical team to become forensic linguistic experts.
Cost effective solution delivered on time for September 2007 and to budget. The solution was designed, tested and delivered on time and within UCAS' budget.