Lessons Learned Developing and Using a Machine Learning Model to Automatically Transcribe 2.3 Million Handwritten Occupation Codes
Machine learning approaches achieve high accuracy for text recognition and are therefore increasingly used for the transcription of handwritten historical sources. However, using machine learning in production requires a streamlined end-to-end pipeline that scales to the dataset size and a model tha...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
International Institute of Social History
2022-01-01
|
| Series: | Historical Life Course Studies |
| Subjects: | |
| Online Access: | https://hlcs.nl/article/view/11331 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832570143208636416 |
|---|---|
| author | Bjørn-Richard Pedersen Einar Holsbø Trygve Andersen Nikita Shvetsov Johan Ravn Hilde Leikny Sommerseth Lars Ailo Bongo |
| author_facet | Bjørn-Richard Pedersen Einar Holsbø Trygve Andersen Nikita Shvetsov Johan Ravn Hilde Leikny Sommerseth Lars Ailo Bongo |
| author_sort | Bjørn-Richard Pedersen |
| collection | DOAJ |
| description | Machine learning approaches achieve high accuracy for text recognition and are therefore increasingly used for the transcription of handwritten historical sources. However, using machine learning in production requires a streamlined end-to-end pipeline that scales to the dataset size and a model that achieves high accuracy with few manual transcriptions. The correctness of the model results must also be verified. This paper describes our lessons learned developing, tuning and using the Occode end-to-end machine learning pipeline for transcribing 2.3 million handwritten occupation codes from the Norwegian 1950 population census. We achieve an accuracy of 97% for the automatically transcribed codes, and we send 3% of the codes for manual verification . We verify that the occupation code distribution found in our results matches the distribution found in our training data, which should be representative for the census as a whole. We believe our approach and lessons learned may be useful for other transcription projects that plan to use machine learning in production. The source code is available at https://github.com/uit-hdl/rhd-codes. |
| format | Article |
| id | doaj-art-7b9631dad0dc480f99c04cbe265b4f3a |
| institution | Kabale University |
| issn | 2352-6343 |
| language | English |
| publishDate | 2022-01-01 |
| publisher | International Institute of Social History |
| record_format | Article |
| series | Historical Life Course Studies |
| spelling | doaj-art-7b9631dad0dc480f99c04cbe265b4f3a2025-02-02T16:24:34ZengInternational Institute of Social HistoryHistorical Life Course Studies2352-63432022-01-011210.51964/hlcs11331Lessons Learned Developing and Using a Machine Learning Model to Automatically Transcribe 2.3 Million Handwritten Occupation CodesBjørn-Richard Pedersen0Einar Holsbø1Trygve Andersen2Nikita Shvetsov3Johan Ravn4Hilde Leikny Sommerseth5Lars Ailo Bongo6Norwegian Historical Data Centre, UiT The Arctic University of NorwayDepartment of Computer Science, UiT The Arctic University of NorwayNorwegian Historical Data Centre, UiT The Arctic University of NorwayDepartment of Computer Science, UiT The Arctic University of NorwayMedsensio AS, Tromsø, NorwayNorwegian Historical Data Centre, UiT The Arctic University of NorwayDepartment of Computer Science, UiT The Arctic University of NorwayMachine learning approaches achieve high accuracy for text recognition and are therefore increasingly used for the transcription of handwritten historical sources. However, using machine learning in production requires a streamlined end-to-end pipeline that scales to the dataset size and a model that achieves high accuracy with few manual transcriptions. The correctness of the model results must also be verified. This paper describes our lessons learned developing, tuning and using the Occode end-to-end machine learning pipeline for transcribing 2.3 million handwritten occupation codes from the Norwegian 1950 population census. We achieve an accuracy of 97% for the automatically transcribed codes, and we send 3% of the codes for manual verification . We verify that the occupation code distribution found in our results matches the distribution found in our training data, which should be representative for the census as a whole. We believe our approach and lessons learned may be useful for other transcription projects that plan to use machine learning in production. The source code is available at https://github.com/uit-hdl/rhd-codes.https://hlcs.nl/article/view/11331Machine learningHistorical dataPipelineCensus1950Norway |
| spellingShingle | Bjørn-Richard Pedersen Einar Holsbø Trygve Andersen Nikita Shvetsov Johan Ravn Hilde Leikny Sommerseth Lars Ailo Bongo Lessons Learned Developing and Using a Machine Learning Model to Automatically Transcribe 2.3 Million Handwritten Occupation Codes Historical Life Course Studies Machine learning Historical data Pipeline Census 1950 Norway |
| title | Lessons Learned Developing and Using a Machine Learning Model to Automatically Transcribe 2.3 Million Handwritten Occupation Codes |
| title_full | Lessons Learned Developing and Using a Machine Learning Model to Automatically Transcribe 2.3 Million Handwritten Occupation Codes |
| title_fullStr | Lessons Learned Developing and Using a Machine Learning Model to Automatically Transcribe 2.3 Million Handwritten Occupation Codes |
| title_full_unstemmed | Lessons Learned Developing and Using a Machine Learning Model to Automatically Transcribe 2.3 Million Handwritten Occupation Codes |
| title_short | Lessons Learned Developing and Using a Machine Learning Model to Automatically Transcribe 2.3 Million Handwritten Occupation Codes |
| title_sort | lessons learned developing and using a machine learning model to automatically transcribe 2 3 million handwritten occupation codes |
| topic | Machine learning Historical data Pipeline Census 1950 Norway |
| url | https://hlcs.nl/article/view/11331 |
| work_keys_str_mv | AT bjørnrichardpedersen lessonslearneddevelopingandusingamachinelearningmodeltoautomaticallytranscribe23millionhandwrittenoccupationcodes AT einarholsbø lessonslearneddevelopingandusingamachinelearningmodeltoautomaticallytranscribe23millionhandwrittenoccupationcodes AT trygveandersen lessonslearneddevelopingandusingamachinelearningmodeltoautomaticallytranscribe23millionhandwrittenoccupationcodes AT nikitashvetsov lessonslearneddevelopingandusingamachinelearningmodeltoautomaticallytranscribe23millionhandwrittenoccupationcodes AT johanravn lessonslearneddevelopingandusingamachinelearningmodeltoautomaticallytranscribe23millionhandwrittenoccupationcodes AT hildeleiknysommerseth lessonslearneddevelopingandusingamachinelearningmodeltoautomaticallytranscribe23millionhandwrittenoccupationcodes AT larsailobongo lessonslearneddevelopingandusingamachinelearningmodeltoautomaticallytranscribe23millionhandwrittenoccupationcodes |