Overall first scan of usefulness
In order to get a first overview of useful this feature in progress might be, I decided to go right to what I suspected would be a case with high likelihood of problems: A child who died at or shortly after birth too quickly to christened https://www.familysearch.org/tree/person/details/LYW6-H8W :
Starting at the top of the Score Details and going down:
Person data complete
- This person is missing given name... This infant was never christened. Never given a name and will never have one. Stars: 0
- The 23 Feb 1873 birth.... Yep. For some reason we didn't put in the family's full residence. Stars: 5
- The 23 Feb 1873 death... Again neglected to enter the full residence. Stars: 5
- The Burial conclusion is missing... Infant deaths in this place and time never had burial records. Of course the program would have to be adjusted for each particular set of circumstances which would not be practical. Stars: 3
Person data tagged
- This statement really doesn't make sense. Stars: 2
Person Data is Consistent with tagged...
- There are no indexed sources.... Yes there are. Of the four sources on the record, two of them are indexed sources. All four sources are tagged to both the Name and Death. Stars: 0
Person Data Conflict Free
No issues identified. But if there are no recognized indexed sources there is nothing to compare! If no comparison is possible does that really mean there are no conflicts? Stars: 1
Overall usefulness score: 2.5 stars.
Adding the farm name to the residence and tagging all the sources to all the information resulted in the following:
This did raise the overall score to 0.93. It can't ever be any higher due to the fact the child does not have a name and does not have a burial.
I find it a bit strange that adding a second indexed record tag cleared the flag that there were none.
My main concern here is that people will get annoyed by not being able to get a "perfect" score or by repeatedly seeing people with non-existent problems but low scores. Would it be possible to add a note field for users to explain why a flagged problem is not a problem and should be left alone?
의견
-
We do need to communicate better that we are not looking for perfection. The algorithm is trying to highlight common, easily knowable issues and things that should be true 90% of the time or more. There will be valid exceptions. We have talked about making issues dismissible, but that is a long way off.
Person Data Conflict Free perhaps should explain more. It doesn't look at sources or care about them. It looks at things like 'were both your parents alive when you were born', 'did you have a statically unlikely number of children for the time and place you lived', 'are birth, marriage, death, and burial in the correct order', 'did your life events happen in distant unrelated places', etc.
0 -
@Gordon Collett thank you for your feedback. Several updates have occurred since this post. And I believe the issues have been taken care of. Please let us know what you think.
0
