TakeLab built a system for normalizing postal addresses. While at first glance this might seem simple, in a morphologically complex language such as Croatian the task is not trivial. The system is still in use by Konzum d.o.o. for storing customer addresses in a standardized format.
People enter their addresses in many various forms, which causes problems with managing and analyzing address data. While at first it may seem simple, the task is not trivial. This problem is particularly prominent in morpohologically rich languages such as Croatian. The goal of this project was an address normalization system which reduces all diferent spellings of an address to a single canonical form. The system was developed for Konzum d.o.o to normalize addresses entered into a form manually by their customers.
Participants: Konzum d.o.o., TakeLab FER
Duration: 6 months