The challenge

Rural development project applications submitted to the Hyrrä system were keyworded by hand, which made the results open to interpretation and inconsistent. The Finnish Food Authority needed a more uniform way to make the themes of the applications visible and comparable.

What Headai built

Headai structured a dataset of 5,585 project applications and built an ontology of 84 keywords from the authority's own data. For each application, the solution returns machine-suggested keywords, a comparison with earlier manual keywording and new theme keyword options, using Headai Graphmind, semantic analytics and a dynamic language model.

Headai acted as the authority's data and AI partner in structuring the content of project applications. The work showed how an authority's own text data can be turned into consistent, transparent and analysable information. Unlike general-purpose generative AI, the solution gives the same result for the same application every time, stays within the agreed ontology and shows the reasoning behind each keyword, which is essential for consistent, auditable processing in public administration. The same approach can be extended with Headai's own large data repositories, such as investments, research and news, to place funded projects in a wider context.

Questions it answers

• What themes does this project application cover?

• How do the machine-suggested keywords compare with earlier manual keywording?

• Which relevant themes are missing from the current keywords?

• How can applications be compared consistently across the whole portfolio?

• Which new theme keywords could better describe the applications?

Impact

Keywording became more consistent and transparent, making application themes easier to compare and analyse across the portfolio.

The machine produced an average of 4.9 keywords per application, compared with 3.9 by human keywording. 99.5% of applications received useful new keyword suggestions, and over 90% received at least three new, precise theme keywords. Targeted fine-tuning improved the result by a further 0.9 keywords per application.

"Artificial intelligence offers support for smoother and more uniform processing of project applications and the preparation of applications. Headai clearly understood the customer's background and needs, which is not a given."
– Tuomas Metsäniemi, Network Expert, Finnish Food Authority

Explore the story

Headai: Consistency as the strength of AI in keywording