Food losses and waste data from non-conventional sources
To continuously update the Food Loss and Waste (FLW) data base, the DataLab implemented a strategy that scrapes the World Wide Web to automatically identify openly accessible reports and studies measuring food loss and waste across food products, stages of the value chain, and geographical areas.
In October 2019, more than 480 publications and reports from various sources (e.g., subnational reports, academic studies, and reports from national and international organizations such as the World Bank, GIZ, FAO, IFPRI, and other sources), which have produced more than 20 thousand data points, were included.
Once the documents were downloaded, specific NLP procedures were applied, to identify patterns like:
These are furtherly validated by a simple human intervention, that has to accept (or refuse) the automatic findings.
The resulting values update the FLW data base, where data can be queried, downloaded, and plotted in an interactive and structured way. The database can be used by anyone who wishes to know more about food losses and waste.