SEXTON RECORDS

@Ashlee C. Aren't we supposed to take into consideration the type of document when doing Name Reviews? Example: Sexton , yes it could be a name of a person, a place or a thing. I saw Sexton as the name of a mining claim ie Sexton, Sexton @2 etc. Sexton could also be a title; ie a gravedigger or a person who looks after a church/chapel. Just my opinion, it matters😎

@Ashlee C. Do you anticipate that a further level of review will be required once AI has produced draft metadata from the full text transcription that the Quick and Full name reviews are helping with?

Thanks @Ashlee C., but in that case how will the dates, places, relationships, and other metadata (occupations etc.) be reviewed, or the split of an image into multiple records, e.g. on directories or censuses? I have read a lot of blog entries and research wiki stuff about FS computer aided indexing, as you suggested previously, but I still don't understand how this all hangs together.

Thank you @Ashlee C., that would be really helpful.

@Sam Sulser could you or the team possibly pick this up & get some clarity from the engineers for us?

The existing materials (as linked to in my most recent comment above) really only talk about the Spanish collections that have been AI-indexed already, and they don't cover the relationship between Computer Aided/Automated/Assisted Indexing and Get Involved at all.

It is very unclear to me what the Get Involved reviews are responsible for.

a) What Image Collections and record types are involved?

b) Are the Name Reviews just helping train AI for future metadata extraction? Or are they reviewing Name metadata that AI has managed to extract from the relevant images, and that will at some point turn up in a Record Collection, tweaked in line with the Get Involved review output? Or something in between?

c) Will Get Involved eventually move on to Places, Dates, Relationships, and the other necessary metadata?

d) Or, perhaps, is Get Involved just aimed at improving Full Text indexing, not Record Collection indexing at all?

It would be great if the CAI documentation in the Research Wiki could be updated to cover these matters, and (more generally) if it were clear to those Getting Involved exactly what they are helping with, which I honestly don't think is currently the case.

Thanks a lot.

(If I am missing something and other people fully understand this, please chip in!)

@Sam Sulser and @MandyShaw1 Here's another possible answer, especially for names like Sexton, etc. In Full Name Review, there needs to be an instruction to read/scan the entire part of the document or at least the entire sentence in which the 'name/word' appears — things like column headings or sentences make a big difference in whether a 'word' is actually a 'name' or is it a title ie Sexton or Elder, etc.