RootsTech 2020 in Salt Lake City saw several announcements about FamilySearch.
|What's new on FamilySearch session|
The first is that FamilySearch have a big problem: they are filming new resources much faster than they are indexing them, so the backlog is getting greater and greater. They currently have 18 petabytes of records digitised (a petabyte is one thousand million million), which is 72 times bigger than the Library of Congress. Even with crowd-sourcing using their 310,000 indexers who add 2 million new names per day, the gap between the number of images captured and the number indexed is still growing. The problem has been exacerbated by the long time it was taking from photographing a record to getting it online (on average 249 days) as there were lots of things to be checked, metadata to develop and so on.
The really exciting news is that they have developed new processes which means it now takes only 24 hours to get new images online. These are, of course, not indexed images. To tackle the indexing bottleneck, FamilySearch is working on having the computer automatically read and index the text, along with relationships mentioned in the record. This is a process called Optical Character Recognition (OCR). FamilySearch are also working on computer handwriting recognition. This is progressing, but is not perfect, and neither is OCR. But now indexers and other users have the opportunity to correct incorrectly indexed entries. Not only does this make the record easier to find, but it teaches the computer about what is correct and what isn’t.
And don’t make the mistake of assuming the records being filmed and indexed are mainly American ones. More of the recent ones are European, but there are also records from South Africa, Asia and South America and even the Republic of the Congo. To find out the what is being added, click on “View the most recently added images” at https://www.familysearch.org/records/images/, or enter a place of interest in the search box on that web page.
These moves will allow for rapid access to content, but indexing volunteers are still needed to index records which will serve to train the computers to get smarter and better at reading the records themselves.
Other announcement made at RootsTech relate to the FamilySearch Family Tree. Below is an interview with Ron Tanner from FamilySearch about the tree.
In summary, new relationship tags have been added. As well as being able to enter a marriage, divorce or annulment for a couple, the couple relationship can be tagged as “Common Law Marriage” or “Lived Together”. You can also add a relationship fact to say that this couple never had any children.
A key aim for the FamilySearch Family Tree is to reduce the poor changes to the tree. If someone attempts to add a child to a couple that is listed as never having children, the system will generate a warning. Part of the strategy to stop these poor changes is to focus on “Hints” and “Possible Duplicates”, so that FamilySearch only shows those it things it thinks are perfect, not those it thinks “might be possible, but might not be correct”.
There is also a new merge function. This will result in a multi-stage merge.
- Decide what to change
- Give a reason and finish.
The process will give warnings if there are data problems, such as dates (e.g. date of birth) between the potential merge more than about 5 years different, birth and death more than 25 miles apart, and anything else that will create a data problem, such as having someone marry their own parent.
A bit of fun that has been added to FamilySearch is the integration of some of the features from the Discovery Centers into the online FamilySearch site. These can be accessed from a new menu group at the top of the page entitled “Activities”. This is only available through the web version, not through the App, though it can be used on a mobile device via a browser.
“My famous relatives” will be added to FamilySearch later this year.