





Archival research can be very time consuming. Sometimes researchers need to look through hundreds of files in the hope of identifying broad patterns or specific pieces of information related to their topics of inquiry. Further, increasing numbers of archival researchers are not trained to read cursive writing. The transcripts that will result from this Zooniverse project will make a big difference because they will help us to make these selected items from Arnold’s collection full-text searchable. This means that future researchers will be able to search keywords (like places, peoples' names and other items of interest), helping to increase their search accuracy, efficiency and scope.
Scholars, students and other interested people from all over the world access the collections of the Dr. John Archer Library and Archives. Thank you for helping us to help them discover the information contained in this rich collection.
We know that there are many worthwhile projects in need of Zooniverse volunteers. We hope that our use of an AI tool to draft the initial transcripts will spare human volunteers from the need to start from scratch on this project. However, while HTRflow is very helpful, handwritten text recognition technology is not perfect. The quality of the transcription can vary depending on factors such as fading ink, condition of the paper, and the handwriting style, and errors like misread letters and words can occur. A human reviewer is essential to catch these mistakes and ensure that the final transcripts are accurate and reliable. Thanks to everyone who is willing to invest their “people power” into helping us cross-check the machine-generated transcripts!
No, we are not using any data generated through this project to train AI models. Data generated in this project are being reviewed by humans, not by AI tools. For more detail on our transcription process, including how HTR differs from OCR, see our Research page.
No, currently there isn't a way to save a draft and return to it. Transcripts also don't save automatically. If you're not able to finish transcribing a page, we encourage you to submit what you've finished.
Do you best to transcribe what you see. If it's unclear or illegible, put the text within unclear tags (i.e., [[unclear]][[/unclear]]). If you'd rather work with a new page, refresh your browser and a different letter should appear.
Every transcript is going to be reviewed by at least three people, and a member of our team will review everyone's transcripts to see where there are differences. If you're not sure what a particular word or phrase is, use the unclear tags (i.e., [[unclear]][[/unclear]]). But, if you forget to use the tags or make some other error, it's not a big deal.
If you'd like a different letter, hit refresh in your browser and a new item will appear. You can also choose a different workflow to try a letter from a new correspondent.
No! If you'd rather delete the AI-generated text and transcribe the letter without it, please feel free!