genealodzy.pl

Hello –

I recently started looking at the scanned documents (Skanoteka) that are linked to some of my ancestor’s indexed records in Geneteka. Many of the documents are in hand-written Russian which I am ignorant of. With the help of Transkribus I was able to get a somewhat useful printed transcription of the hand-written text, which I could then copy into Google translate to get a rough translation of the text. The translation was pretty rough, I suspect, because the model of Russian handwriting available in Transkribus that I used was from a modern diary, so some of the orthography and vocabulary of a 19th century Parish Document was not exactly fitted to the model they had available. Even with these issues, I was able to learn a few things from the rough translation.

If a better model was created in Transkribus based on typical documents from Skanoteka records, I was wondering:

• Could Transkribus be (more) helpful to those of you who regularly transcribe hand-written Russian documents?

• Would researchers like me, who are ignorant of Russian, be able to get useful, if rough, translations by themselves?

I was referred to Transkribus by a link at your https://genealodzy.pl/ site, but I am unsure to what extent you all use it in your translation work. And I am wondering, if any of you hand-written Russian reader/translators might be willing to try to create a useful model for those of us trying to understand the content of these documents?

I would be willing to help in this pilot project, but since I do not know Russian, I could only try to help with understanding Transkribus and their processes.

From what I gather from looking at some of the Transkribus videos (available at https://www.youtube.com/@transkribus/videos ) about 20 pages for records would need to be put into a Transkribus collection file and initially transcribed using the imperfect Russian model now available. The real work would begin – the print transcription would need to be manually corrected by the Russian expert, so that the print transcription exactly matches the hand-written original. Then this corrected version could be used as a new model for better transcriptions of subsequent Skanoteka documents.

So, if anyone knows the answers to my questions above, or if you are a Russian translator who would like to try create a useful model in Transkribus, let me know.

Thanks (and thank you all for creating the GREAT resources at this site!)
- Bob Schroeder

I know results can be wonky, I copy and paste the text it generates to google translate and also ChatGPT. Since I usually know what I'm looking for phonetically I can usually decipher it. Also, this site has been very helpful as well: https://stevemorse.org/russian/cyrprint ... font=print

I have been able to figure out tons of documents this way. They aren't perfect but it's led me to a lot of ancestors.

Hi Karen,

I do the same with the Transkribus results in Google Translate. I'm usually looking for dates and names so it works for that level of sophistication. It has allowed me to verify and refine a lot of the relatives I've found in Geneteka or MyHeritage.

-Bob