This project involved investigation of how natural language understanding and generation technologies could be used to assist a report writer in generating a report for a large test, involving masses of data of heterogeneous sorts.