From the University of Maryland:
University of Maryland faculty and graduate students in computer science and economics, with a colleague from UCLA, have created the largest national database of food safety inspection information, InspectionRepo.
In the U.S., such inspections are done by local public health departments, which can take different approaches to conducting, coding and reporting inspection data. Using this unique new automated database, food service businesses and consumers can monitor and compare food safety practices from outlets across the nation.
The national database was developed by UMD Professor of Computer Science Ben Bederson, UMD Professor of Economics Ginger Jin, UCLA Associate Professor of Business Management Philip Leslie, new Ph.D. graduate Alexander Quinn (computer science) and UMD Ph.D. graduate student Ben Zou (economics).
According to Bederson, who also is UMD’s Associate Provost of Learning Initiatives and Executive Director of its Teaching and Learning Transformation Center, the team’s database uses data robots to automatically collect data from local government websites, and represents a huge leap from local and state databases that are built using manually-collected and sometimes poorly correlated data, and which can easily miss the big picture and have little impact on compliance actions.
“Our data robots cover a large number of local jurisdictions across the U.S., continuously detecting new data posted by each jurisdiction, and integrating them into a single, standardized, and cumulative database,” [aka web-scraping] Bederson said, noting that the result is a database that is cost-effective, robust and scalable compared to manual alternatives.
Another difficulty was developing normalization algorithms to compare data across jurisdictions where the data is very different. For some web pages, the team had to write custom ‘scrapers’ to get the data, and for others they had to interpret already available databases.
The researchers also developed analytical tools that can be used to compare inspection outcomes across localities and states, and across chain and individual food outlets, such as restaurants, cafes, convenience stores, and grocery markets. This can improve inspection efficiency and promote retailer compliance, resulting in a decrease in food-borne illnesses, according to Bederson.
The team has created a regulatory data analytics company, Hazel Analytics, which according to Bederson is a direct outgrowth of their academic collaboration around food safety inspection data funded by the Sloan Foundation.
For non-commercial use, the database is publicly available at InspectionRepo.com at no cost.
Read the Complete U. of MD Announcement About InspectionRepo.