Through this proof of concept project, a range of handwritten documents were run through Clio and through a range of checks and balances, the results of the AI transcription far exceeded the expectations that were set by QSA, with around 80% of the transcriptions being above average or better.

The resultant transcriptions were then passed on to the Queensland State Archives team, where they were able to integrate the results into their existing crowd-sourcing program to further check and review the results.  This reduced the time it took to transcribe the archives, and made it possible to include more materials in the launch of the Frontier Wars material.

You can read more about this project in our blog.

The post Transcribing Archives with Artificial Intelligence appeared first on Gaia Resources.

A number of these scanned forms were run through Clio in two runs – the first run saw the system return a lower accuracy of 73%, but after working through the forms and tuning the models in the AI system, we were able to get it to between 93% and 99% in the second run.

This approach of using a targeted and specific AI tool helped the Department of Environment, Parks and Water Security to capture information from field sheets in a much faster, less error-prone method than the current manual system that they previously employed.

You can read more about the project on our blog.

The post Transcribing Field Notes with Artificial Intelligence appeared first on Gaia Resources.