AI-supported improvement of data quality in NOGA code assignment

Improving data quality with AI: Automated classification of NOGA codes

A company from the sector Economic development and location development It manages a large amount of company data, including for statistical evaluations and economic analyses. A key problem: The automatic or manual assignment of NOGA codes was often inconsistent, error-prone, and opaque.This is especially true for new or unusual company names and job descriptions. To address this challenge, a new approach has been developed to improve data quality using AI.

Solution with Comitas: AI-supported classification based on GPT

Comitas developed a prototype together with the client, which used GPT-supported text analysis:

  • Free text descriptions of companies analyzed
  • Recognizes industry keywords and links them to official NOGA categories
  • Provides suggestions with confidence values ​​for selection
  • Automatically flags inconsistent or missing entries.

The solution was embedded into the existing database environment and made usable via a simple web interface.

Result:

  • Improving data consistency and significantly fewer manual corrections
  • Transparent, documented suggestion logic for specialist departments
  • Accelerated data maintenance and real-time quality monitoring

"The AI ​​recognizes industry texts that are relevant for People difficult to categorize would be – and thus save us many manual assignments every day.” – Data controller (anonymous)