How might we clean, categorise and analyse structured and unstructured data with minimal man-in-the-loop (manual work), in order to gain useful insights?

Hundreds of reports are received in the form of free text on a regular basis. Manual cleaning, categorisation and analysis of the reports are performed by a few individuals for submissions and the archives. The data-cleaning and analysis process is tedious and resource-consuming, as the individuals need to be trained to understand the reports, classify them, and also make sound analysis if and when required.

The solution should easily be executed by a team of 2 staffers (working in shifts) with minimal training required. The task should also be easy to complete, with a target rate of 100 reports within 5 minutes and an accuracy of 95%.

The solution can be introduced in phases. The Minimal Viable Product (MVP) should automate the categorisation of report, executed on demand by existing staff. The MVP should minimally demonstrate 95% accuracy, in order for the solution to gain support for future development.

Selected solution providers will get a chance to run a pilot, fully funded by MINDEF.

Solution providers are encouraged to include a summary quotation along with their submission.
Please refer to the Terms and Conditions and FAQ for further details.