Are we allowed to use ? and * when correcting names in the Family or Names review of the 1950 Census?
Comments
-
Are we allowed to use ? and * when correcting names in the Family or Names review of the 1950 Census?
2 -
Good question @John Empoliti - I doubt AI knows about asterisks for unreadable -- I also found a line where AI showed "None" for a name when the census-taker had written "No one home" for a particular apt. # address --- I reviewed it as a transcription error but I don't know how that will work once the document is online -- let's just pray that internet/political powers let us keep the original census visible instead of destroying it leaving only the AI indexing behind😎
1 -
I was thinking about the situation in which I consider a letter or string of letters is unreadable - notwithstanding that the AI has rendered an opinion. Thus the need for "?" or "*" even in these AI-derived indexes. I know we're supposed to be training the AI, but we're also delivering the result that FamilySearch will publish, and their search engine knows about "?" and "*." We could choose "Unsure" in those cases, but I wondered about using our normal practice in cases like that.
For the "No one home," "Vacant," and similar AI results, I select "Transcription Error>Not a Name "
0 -
I think we should try using ? and * for letters we cannot read in a name or place and see what happens!
1 -
For regular reviewing, using ? and * is correct, but not for the 1950 census. John Empoliti is correct in using UNSURE. Who knows what the HRT is programmed to do when we click on UNSURE. It's like deleting "NO ONE HOME" instead of marking it as NOT A NAME. We need to follow the rules as directed until told otherwise.
1 -
Yes, it's fine to use the ? and * when reviewing the names for the census.
2 -
Often, when a name listed in the census is unreadable or doesn't look quite right, I research the inviduals. For example, I search newspapers.com and findagrave.com to ensure that odd-looking names really do exist. I search for variations, as needed, along with birth states and approximate birth years. I check for relatives listed with them, and so on... When I do so, I can often identify these individuals quite confidently, and I put their "verified" names in the edit box.
Does this practice interfere with AI-training that we are supposedly doing while we review names?... And could the "verified" names get removed by later reviews and edits done by those who haven't researched them and are merely ensuring correct transcriptions?
1 -
The important thing for the AI is to type what you see. It doesn't hurt to do a little investigation on the side as it sometimes helps with unusual names like you pointed out. Just don't make corrections to what the census taker wrote. When the index is available to be searched, then census taker corrections can be made. Each edit or correction is kept in a history and all of them will be searchable and viewable. Hope that answers your question! Sam ☺️
0
