Over 1850 man hours per month are spent answering compliance related questions such as “What is the gift limit for UK?” or “What is the entertainment policy for USA?”. The Compliance department usually has a manual process to follow this and has to identify and read multiple PDFs to answer this. DataSpartan was tasked with automating certain aspects of this process.
A customised search engine was built where PDF documents were categorised according to importance by the term frequency – inverse document frequency (TF-IDF) metric. This would rank the PDFs based on the initial search by the user and present the PDFs in form of relevancy.
A graphical user interface was built and deployed onto the client servers and the system is expected to save 115 man hours per month for compliance issues (6.2% improvement).