Stop the display of "Similar Records" if they are already attached to that profile
I believe this issue has been raised before, but looking (as an example) at the page at https://www.familysearch.org/tree/person/sources/LY71-7P5 (see below) has illustrated good reason for these Similar Records not to appear against the profiles of individuals to whom they are already attached. I just cannot see the point of this - okay if they had been attached (either correctly or in error) to another individual in Family Tree, but why do we have to be shown something that is clearly found through scrolling through the Sources section of LY71-7P5? If other users do find these details are useful, please tell me what I am missing, in believing them be of no added value whatsoever!
Comments
-
Perhaps this is the thread you mean, @Paul W: https://community.familysearch.org/en/discussion/comment/530250#Comment_530250
1 -
There is another situation in which the inability to remove items from that list (or remove that list, period) is even worse of a problem: when the record is for a different, but similar, person. It seems like this part of the hinting system is specifically designed to encourage conflations.
2 -
One good thing about leaving the hints in: So people will be more accurate is only linking those hints to the family that they Know are correct. I have found some who just add any hint they get making it more difficult to filter out good facts from speculation. You most likely do a very good job of checking out your hints. Unfortunately everyone does not.
0 -
@WKathleenM4 But would the people who don't take the time to factcheck pay attention to the hints, if they won't even take the time to read the details of records? And the similar records need factchecking as well. In my family, there was a Samuel who married a Sarah, who had a cousin in the same town named Samuel who married a different Sarah. I have little doubt that the Similar Records algorithm gets the two confused. That may be an extreme example, but surely it doesn't even need to be that close to throw things off, particularly when someone has a common name.
2 -
It doesn't even need to be a particularly common name. Locally common will do, or even just common in the family. Combined with the hinting system's general tendency to ignore women's surnames wholesale, and to find matches between completely-different surnames, I really, really do not see any utility to the whole "similar records" section.
Well, OK, that's not entirely true. If there are many indexed records from a time and place, but a particular entry has no similar records, it can be a clue that possibly, something was misindexed. But that's something to be investigated on not-yet-attached records, not on already-attached ones (which was presumably done based on checking the actual image).
1 -
It doesn't even need to be a particularly common name. Locally common will do, or even just common in the family. Combined with the hinting system's general tendency to ignore women's surnames wholesale,
This week I had a hint (dismissed quickly) for a male baptism. First and middle names matched. Date of birth/baptism matched. Surname was not a match, not even close. I was researching the Michalak family. When I found the baptism, the hint algorithm suggested a good match would be a child with the surname Binias.
0 -
This week I had a hint (dismissed quickly) for a male baptism. First and middle names matched. Date of birth/baptism matched. Surname was not a match, not even close. I was researching the Michalak family. When I found the baptism, the hint algorithm suggested a good match would be a child with the surname Binias.
Ehh, I can see the logic there. If everything else matched exactly but the surname, it's possible the record exists under two surnames (which I've seen when a child is born out of wedlock), or if the name was severely mistranscribed on one derivative source (which I've seen a ton on early 20th C. records for Italian immigrants.
2 -
Not in this case. Binias is nowhere close to Michalak, even if the cat was using the keyboard.
0 -
(We're talking about Similar Records, not Hints, right? I'm only talking about having them show on the Similar Records list.)
Those names aren't as different from a purely visual standpoint as they are lingually. Either name being scanned from smudged, faded old handwritten document could look vaguely like ∦i∩∣a≲ to an OCR. Here's a record for Iiattarara, which is actually Quartararo -- vaguely similar to the eyes, very different to the ears. Here's one where you can tell the census taker botched the name spoken to them and the transcription messed it up further: Curshus Joecone, which I think is supposed to be Accursio Iaccino.
There are definitely cases where it's worth looking at context instead of fixating on names alone. I've seen a ton of death records that had an parent's name wrong (usually mother's maiden name). The town in Sicily where one of my branches comes from was fairly open policy about giving up unwanted children, and they often changed their surnames in adulthood. There's also a few countries that had dit names or familiar names. Plus we still have tons of records that look like this: ?agurs* Ursy.
I'm glad that there's an option with broader matching that takes context into account. I have found records that I wouldn't have otherwise, opening branches I never would have found. The records that didn't match the current person were often just as helpful -- they made me aware of superficially similar families that might be confused for the one I'm working on, so I can clean up any existing errors, mark them as Not a Match and put up alerts and warnings for others to ignore.
Hiding Similar Records that are already attached to the current person seems like a decent idea to me, though it will result in that list being turned into only records for other people that look similar. If it's being misused, I'd blame the editors misusing it rather than the feature itself, but it'd also be good to have it (and merging) as opt-in features that are turned off by default. I don't want to see them completely removed or significantly changed, except for them fixing the followed-link color and right-click to open when in record view bugs that were raised a year ago.
1 -
@RTorchia, I'm curious: do you actually use the Similar Records list on a profile's Sources page? Or on the Sources panel? For me, it's just a serious annoyance in both of those places. It makes it hard to look at the person's sources, because there's constantly this list in the way. I would much prefer if that list only showed on the index entry's detail page, like the Document Information (inexplicably) does now. (Why does the detail about the source itself not show with the source itself, while the long list of unrelated stuff does?) In other words, I would like to switch the behavior of the Sources page and panel with respect to the index detail page's left-hand and right-hand columns.
2 -
@Julia Szent-Györgyi Yes, frequently, and to good effect, especially when filling out English families from the 17th C.-18th C. There are a lot of records that don't get flagged as hints for whatever reason -- maybe they short Thomas to Thos., have the Latin versions of the names, a minor surname misspelling, etc. I've also found it very useful for finding census reports that don't load as hints for similar reasons.
It's also useful for finding the correct parents already in the system in situations such as someone merging an English family into an American one, mass-merges, sources attached to the wrong people, or when people just ignore locations and grab whatever sources vaguely fit the name they're looking for.
0 -
@RTorchia, I understand the use of the Similar Records list in general, but I was asking specifically about looking at it from a profile (the Sources tab or panel) rather than from an index detail page. Do you actually use those views of the list?
0 -
@Julia Szent-Györgyi Yes, very often, that's what I was referring to when I mentioned applicable sources that don't get included in Hints -- those are found from reviewing what's already attached in Sources. I'd be kind of surprised if an editor that's trying to ensure a profile is complete doesn't go through that list.
0
