Request to correct geographical indexing in Tennessee 1870 and 1880 U.S. Censuses
Tennessee has 95 counties that are subdivided into numerical Civil Districts, which are used as the locations of the 1870 and 1880 U.S. Censuses. For example, Hawkins Co. has Civil Districts 1-17. Because many of these Civil Districts have not been added to the list of standardized place names, many of these records are being incorrectly indexed to the wrong county, often Grainger Co., which apparently does contain Civil Districts in the standardized place name list. This, of course, makes searching in Tennessee much more difficult than it needs to be. Since 1870-1880 is the period when formerly enslaved families were enumerated by name and relation for the first time, it is important that this be corrected. It seems like it might be possible to prepare lists of the Civil Districts for each county and to create the standardized place names en masse.
I also submitted a suggestion (which are apparently currently being withheld) that the place name matching algorithm match hierarchically, rather than "left to right." For example:
https://www.familysearch.org/ark:/61903/1:1:MDDQ-VLF
Event Place | District 12, Grainger, Tennessee, United States |
---|---|
Event Place (Original) | District 12, Hawkins, Tennessee, United States |
With a hierarchical algorithm, "United States" matches, then "Tennessee" matches, then "Hawkins" matches, then "District 12" fails. Therefore, the match would return, "Hawkins, Tennessee, United States" which is a huge improvement over "District 12, Grainger, Tennessee, United States."
Thank you for your consideration of this request.
Answers
-
I also would be interested to know what is happening on Suggest an Idea - I submitted an Idea last Monday, as advised by FS Support (who said I should ignore the warning message that is currently shown), but my Idea still has not popped up on the list.
@Ashlee C. can you help please, with the OP's specific placename report, and also concerning the withheld Ideas? Thanks.
0 -
Unfortunately, it's not just Tennessee. And, it is not an indexing issue.
If you browse through the Community you will find many such reports for nearly every conceivable location. The cause is an automated routine that has sadly corrupted the location data for many records.
1 -
@Anne986 requested that
… the place name matching algorithm match hierarchically, rather than "left to right." …
It is absolutely clear that the standardisation place matching needs to be improved. Totally agree. But I honestly don't think it's as simple as altering from a "left to right" match to a "hierarchical" match (right to left, I guess…)
It has to be more complicated than that, to cope with omission of name parts, insertion of others, spelling variations, etc. I'm guessing that there are patterns that get invoked so that a "match" on parts 1 and 3, and also on part 4 of the name, means something more than just a match on part 2 and on part 4. I put "match" in quotes because there's also variant names that need to be checked, adding another level of messiness.
I totally agree that your suggestion of standardising on "Hawkins, Tennessee, United States" is the correct answer if "District 12, Hawkins, Tennessee, United States" isn't available in the list of standardised places. But I'm not sure that it's as simple as reversing the order of checks. I think we have to believe that the FS programmers can eventually work things through if there are presented with examples like yours.
0 -
Mandy, Áine, and Adrian,
Thank you for your comments. The matching algorithm could go left to right, although it seems less efficient. The main point is to end with a match that is completely correct, albeit, less precise. In the left to right version with the same example, it would match on "District 12" and then fail on "District 12, Hawkins." So the algorithm would next omit the failed "District 12" and search for matches starting with "Hawkins." It would match "Hawkins," then match "Hawkins, Tennessee," then match "Hawkins, Tennessee, United States." Therefore it would return the match, "Hawkins, Tennessee, United States."
0