News

26.11.2020

Judgment books freely available through new online service

The new court records search service of the National Archives of Finland is available to the general public as of today. In the service, you can browse and search for renovated 19th century court records from the Finnish territory. Digitised documents were processed by means of an automatic optical character recognition (OCR) system, Handwritten Text Recognition (HTR), which utilises artificial intelligence.

The renovated court records are one of the largest collections of the National Archives of Finland. The documents saved in the online service are records from registration cases from 1809 to 1870. They involve registrations of titles to properties, guardianship cases and prenuptial agreements. The records from registration cases can be used for genealogical purposes or to trace the ownership of properties, for example.

The court records search service also includes features to facilitate searching, such as search term lists and maps to clarify the division of judicial districts over the years. In addition, there are instructional videos on the website of the National Archives of Finland on how to use the service, which offer new users ideas on how to utilise the court records search.

A test version of the court records search service was made available to users in September, and feedback from the test period was predominantly positive. “Based on the feedback, it seems that there is plenty of demand for the court records search service,” says Maria Kallio, Senior Research Officer at the National Archives of Finland. “One of our goals was to ensure the availability of the service, and the test period confirmed that the service is ready to be made available to the general public.”

OCR is not completely flawless. All of the materials in the service were recognised by the AI, and none of the documents have been edited since. The technology is quickly being developed, however, which means that materials recognised by the AI can be expected to improve in the future. In addition, the plan is to further supplement the materials available in the court records search service with records from actual 19th century legal cases, as well as to expand the chronological period covered. AI to recognise handwriting from the period 1880–1918 is currently being developed.

Based on artificial intelligence

The transcriptions available in the service were created by using the HTR technology (Handwritten Text Recognition), which is based on cognitive artificial intelligence. It is being used as the basis when creating optical character recognition models for transcribed pages, i.e. pages converted into modern text. The National Archives of Finland developed a recognition model specifically for court records from the 19th century, and the materials in the online service were automatically transcribed with this model.

The search service is based on keyword spotting technology that searches for a keyword from probability matrices generated by the HTR model. This means that the keyword is not being searched from the transcribed text but from data generated to support the image. The HTR model issues a probability for each letter of the alphabet based on how certain the model is that a specific part of the image corresponds to a specific letter.

The National Archives of Finland created the OCR model used in the court records search service in a project funded by the EU, READ (Recognition and Enrichment of Archival Documents). The recognition and enrichment of archival documents now continues in the European Cooperative Society READ COOP. In addition to the judgment book materials, the National Archives of Finland utilises the results of the project in a project called Making a Modern Archive, where the technologies developed by the project are being integrated into the digital infrastructure of the National Archives of Finland.

Search Finnish Court Records

Further information:

Maria Kallio, Senior Research Officer, tel. +358 29 533 7194, maria.kallio@arkisto.fi

News >>

Frontpage >>