Extractive text summarization and document classification


Marko Latvanen

Specialist at Population Register Centre (VRK)

The Finnish Population Register Centre (VRK) promotes digitalization in Finland and they’re constantly looking for solutions that enable the optimum build and delivery of e-services, improve cooperation between public administrations and help make internal processes more efficient.

Currently, there are more than 25,000 service descriptions in Suomi.fi (a service provided by VRK) depicting more than hundreds of different services - with more coming in for inspection every day. To reduce the editorial board’s workload and having more organized services, VRK decided to create templates for all the different services listed on Suomi.fi. Digital Workforce developed a natural language processing based solution for VRK that can automatically pick out contextually similar documents from a large pool of service descriptions and extract a summary of them and use those summarizations as a template for future service descriptions. In this webinar, you’ll hear how’s and why’s behind this interesting project.

Marko Latvanen is a specialist at the Population Register Centre (VRK) and he worked with the Suomi.fi services for over ten years, initially as a content editor and later as a researcher and service concept developer. He has a strong interest in ethical AI, integration of AI into natural language content production and management, and semantic interoperability.

Key takeaways from webinar will include:

  • Creating a cheat sheet for description writers by utilizing AI in summarizing documents & creating templates and categorizing
  • Teaching & optimizing the algorithm
  • The challenges of building semantic level natural language processing
  • The results of the project

Watch the recorded webinar:

By submitting this form you agree that we may contact you in a manner as describribed in our privacy policy.