Inflection of Slavic names vs Type What You See
Recently I've noticed a newly indexed Czech Republic church books collection, https://www.familysearch.org/search/collection/1804263
I appreciate this activity a lot as it helped me to find some missing pieces in my family tree.
However, indexing of these records follows the Type What You See rule, which is IMHO suboptimal for Slavic languages.
The typical record:
Jan Hradecký, syn Karla Hradeckého (JH, son of KH)
is indexed as:
Principal: Jan Hradecký
Father: Karla Hradeckého
instead of expected:
Father: Karel Hradecký
The latter is the form that people usually type when searching. When used, that inflected form is not listed in search results as it diverges too much.
I understand it is tricky for non-Slavic speakers to determine non-inflected variants of the names.
Is it OK to override existing indexing? I plan to fix it for records related to several families I maintain.
Thanks,
Jan
Best Answer
-
The same problem applies to many languages. For example, in Latin records, the names of the parents are often in the genitive (possessive) case, and in German records, many women have the feminine suffix "-in" added to their surnames. Granted, the problem is especially bad in Slavic languages, because the inflections affect surnames, and FS's criteria for those are considerably narrower than they are for given names.
I've removed superfluous suffixes from the indexed records for my relatives, but I haven't been at all conscientous about it: after all, the index has already served its purpose, despite the error, because I found the record. But if it makes you feel better, and the index is correctable, then by all means, go ahead and make the corrections.
0
Answers
-
My biggest concern was whether such edits are acceptable or not. Especially seeing that all indexing batches are carefully reviewed before publishing. And now I'll make adjustments without any subsequent approval, where the final variant diverges the form in the image. But I expect if this was unwanted, any post-editing would be disabled.
I was also unsure if indexing is not used for training some custom ML models. I expect for this purpose the original inflected version would be preferred.
OK, thanks for confirming, I'll go ahead and improve items for related families.
0