The Process of Unit Price Extraction from Public Sector Contracts

Tomáš Bruckner, Filip Vencovský
2020 Acta Informatica Pragensia  
Czech government institutions commissioned a research on extracting usual unit prices from public IT contracts to aid future public tender sizing. The goal of the project is to obtain millions of con tracts from the public register, convert them to full text, extract unit prices from the text and publish a pricelist of IT industry manday prices. This paper designs the process and method of price ex traction, demonstrates and evaluates the result on five iterations of extraction and discusses
more » ... on and discusses the experience of two years of project performance. The process is designed as a set of repeatable workflows and specified activity and role description. The method is designed as a combination of automated and manual actions. Due to the format and content variability of involved documents and the low mistake tolerance, the possibility of automated extraction of unit prices from full text contract is limited, and human workforce for validation is crucial.
doi:10.18267/j.aip.139 fatcat:5sdgwe6bgjg67ang3fh3xkj6fe