This project was initiated to deliver a proof-of-concept system architecture to examine documentation for references to Property Flood Resilience (PFR) and score documents depending on the usefulness of the information they provide. These documents mentioned are stored in a variety of network locations and it is not necessarily known which of them contain useful information on PFR.
Work done and value provided:
The AECOM machine learning team delivered a presentation that demonstrates a proof-of-concept of this solution, and a report to summarise the findings. Based on the findings, the machine learning team wrote an algorithm that was using natural language processing and understanding to scan the documents mentioned above. It was provided with existing project information, so that further information (e.g. location) could be appended, which greatly improved efficiency of the process. In addition to this, PowerBI dashboards were built to provide a high-level overview of data gathered in this process As next steps, it was recommended to integrate this algorithm and dashboard with MS Search capabilities once MS Office365 rollout is completed at the Environment Agency.