RLP 135: BYU Record Linking Lab and the 1918 Influenza Pandemic

Nicole (2s):
This is Research Like a Pro episode 135, the BYU Record Linking Lab and the 1918 Influenza Pandemic. Welcome to Research Like a Pro a Genealogy Podcast about taking your research to the next level, hosted by Nicole Dyer and Diana Elder accredited genealogy professional. Diana and Nicole are the mother-daughter team at FamilyLocket.com and the creators of the Amazon bestselling book, The Research Like a Pro a Genealogists Guide. I’m Nicole co-host of the podcast join Diana and me as we discuss how to stay organized, make progress in our research and solve difficult cases.

Nicole (44s):
Let’s go. Hi everyone. And welcome to the show. I’m Nicole Dyer co-host of Research Like a Pro, and I’m here with my mother accredited genealogist, Diana Elder. Hi Diana.

Diana (57s):
Hi, Nicole, how are you doing? What have you been working on?

Nicole (1m 1s):
I’m good. I have been working a little bit on the Daugherty case after we had our last two podcast episodes all about that. I got some great emails from people who descend from the Daugherty Taylor family, and that has been really fun. So I thought I’d give a little update. One of the listeners who heard that episode is part of a Daugherty Facebook group, and I’m part of that group as well. And he posted the link to the podcast in there. And then the author of a Daugherty family book reached out to me and I already had purchased her book last year, and we were talking a little bit more about it. And she reminded me that there’s a whole section in there about James Taylor, who I mentioned a bit who married Sarah Daugherty .

Nicole (1m 43s):
So I went and revisited that chapter in the book and looked at their descendants and they actually have a daughter named Sarah, Sally, Taylor, who married a Patterson and their line hasn’t been traced out, and I wonder if she was maybe the mother of John Robert Dyer, but I don’t have any DNA matches to that line. So I don’t know, but it is something to pursue. The author of the Daugherty book, her name is Donna Hart Burden and she has done a lot of research. Her book is like thousands of pages long, and it has a separate book that’s the index to it. So it’s got a lot of Daughertys in it. And she’s let me know that, that name, that we were wondering about Farnifole is actually in the book a couple of times.

Nicole (2m 23s):
The one that I had mentioned in the podcast was Farnifole Taylor, who was also a witness on the same marriage permission that Peter Daugherty was on. Well, there was another man named Farnifole Green who was also an associate of some of the earlier Daughertys. Apparently he was a clerk who took a list of militia in 1754 in Craven county. Well, several years later there was a, a Farnifold Green Daugherty who seems to be named after the Farnifore Green, but his name was Farnifold with a D at the end. And he was born in 1849. So that name kind of was a popular name. I wouldn’t say popular, but it was a name that was used in Craven county and among the Daugherty descendants.

Diana (3m 9s):
Wow. It’s so exciting what collaboration can do. Isn’t it, sometimes we think we know everything about our families and then we find out there’s all this other research and people that have been working and know so much more so.

Nicole (3m 22s):
Yeah. And one of our listeners sent me some information about a town in Scotland that is similar to the name Farnifore. So she sent me that as Forfar. One of our other listeners is a descendant of James Taylor and Sarah Daugherty and she sent me her GEDmatch kit numbers to run mine against. And none of them were a match. She had two and I ran my eight against them and she ran them actually and found that they did not, but it was interesting to correspond with all these Daugherty people. I love it. It’s great to collaborate and hopefully I’ll be able to figure out the answer to right, the Dyers match this cluster of Daugherty descendants someday.

Diana (4m 6s):
That’s going to be fabulous when you do work so hard on it. But, you know, especially when working with DNA, sometimes it just takes the right people to test and to get the matches that you need. And so eventually, hopefully that will happen. Well, let’s talk about our announcements for the day. We have our study group for 2021, the spring group starting in March, and we have a few spots left in the afternoon and evening groups. So anyone who’s been thinking about joining a study group, this might be the perfect time for you. We love working with individuals, helping them learn how to do the Research Like a Pro process and get lots of feedback and help with their research.

Diana (4m 49s):
And then we’re also excited that we have launched the Research Like a Pro with DNA e-course, and we’ve already had several people sign up for that and are going to work on discovering how to use their DNA test results in a real project. So it’s really fun to see all the objectives that people are creating and I’m excited to see their progress. So if you’re interested in learning how to work with your DNA, check out our new course, and you can find all the information about that on Family Locket and get details, everything that you need to know, if it’s something that you are ready for.

Nicole (5m 29s):
All right, great. We are so excited today to be talking with Joe Price of Brigham Young University. And we first met him at Brigham Young University’s education week, and we were teaching there in the same room as the family history room. So all the classes there were about that topic, and he’s actually an economics professor, which is an interesting connection to family history. So I’m going to let Joe tell us about himself.

Joe Price (5m 57s):
Sure. So, yeah I’m an economics professor at BYU, and about five years ago, I discovered family history as a personal passion, but I didn’t know how to combine it with my work. And then I was at a conference and I bumped into Martha Bailey who’s a professor at Michigan. And she said that she was linking together the entire population of Ohio. And that was the first time I realized that economics could be combined with family history. So I created a lab called the Record Linking Lab at BYU, has about 55 students that work in it and we basically combine economics with machine learning with family history to try to link together a large populations and use it to study migration, mobility, and other like the effects of different economic shocks or something like a 1918 pandemic.

Nicole (6m 39s):
It’s really neat to think about all the connections that you’re making between economics history and family history and tying them all together.

Joe Price (6m 47s):
Yeah. I mean, I think economists were trying to do it just as a machine learning exercise. And then I think family historians have been doing it just as a pure human endeavor. And I really think that we’re going to get a lot of synergy by combining the two approaches, combining the machine learning with the human insight. And so I’m excited to see kind of how that plays out.

Diana (7m 4s):
Well, I am so curious about how you work this with the students because you were there at Brigham Young University. And do you work with quite a large group of students? How does that all come together?

Joe Price (7m 18s):
Yes. I have 55 students that work for me. I have three students that recently graduated that work full-time that help manage a lot of the aspects of the lab and mentor the students, but then students are just grouped into teams. So we have a team that works on computer vision, which is extracting texts from images. I have a team that works on record linking, and then we have a lot of smaller projects. For example, someone approached us about Okinawa. So I found a student from Japan. He was able to work on this project for Okinawa, the same thing with the Czech Republic. And one of the neat things about BYU is we have so many students that are either from their countries or serving a mission in other countries and speak the language and build the culture that I think BYU is almost a perfect place for this type of lab.

Diana (7m 58s):
Oh, that is for sure. Do these students come with an interest in family history or do they develop a interest as they’re working on this? What do you see with that?

Joe Price (8m 9s):
Yeah, the application definitely has a section asking them about their experience with family history and anyone that’s had the family history class here on campus, or has been involved in index scene or then a consultant. Those, I always kind of flag as students that I would love to have come into the lab, but also I like, I love taking a student that’s never done family history before and planting a bit of love in their heart because I really think part of what the lab’s going to do is create people that are going to go out and do family history, maybe a new ways and engage people in new ways.

Diana (8m 39s):
Wow. That’s really exciting. And when you were talking about, you know, marrying economics with family history, it really made me think about the census because that’s why the census was created in the first place was kind of economics, you know, the government wanting just to know how much money they’re going to have to pay out and pensions or how they’re going to collect as taxes. And so we don’t really talk about economics a lot and family history and making those connections, but I think it’s always there. So this is really kind of a fun topic to discuss how your lab is working.

Joe Price (9m 10s):
Yeah. And you can see your point exactly. In the questions that show up in the census over time, you know, they started adding occupation and then 1940, we added income and education. And so you see that the census of all four, maybe just counting the population to understand the population and using it to make economic decisions as a country.

Diana (9m 30s):
Yeah, absolutely.

Nicole (9m 30s):
I’m curious if you can just tell us a little bit more about what the record linking lab actually does.

Joe Price (9m 36s):
Yeah. We do three things. We do computer vision, which is extracting texts from images. We do natural language processing, which is turning that text into usable data, you know, tagging names, dates, and places. And then we do record linking, which is then linking someone from one record to another. And all of these are with the purpose of growing the family tree. And so like here would be one example of a project that combines all three. We gathered records of students who had gone to Harvard. So we went to the Harvard archives to pictures. And then from those pictures we converted into text. We then link the text to census records. And now we’ve been able to write a paper about the impact of your roommates in college, on your long run outcomes and also the impact of social clubs on long run outcomes.

Joe Price (10m 19s):
So that’s a case in which, you know, we were able to answer an economic question by gathering new data, turning it into usable data, linking it with other records. That’s kind of the core of what we do more broadly right now. We’re trying to link together. Everyone that lived in the United States from 1850 to 1940 across census records, we’re creating a pipeline where we could take like the Czech census, the Congo census, and automatically index it and then turn it into usable data. We have some new indexing experiences designed particularly to help younger children and get involved in indexing. And then we put in a lot of focus on helping increase the coverage of the tree for African-American families.

Nicole (10m 58s):
That is so cool. We’ll talk a little more later in the podcast about how you work with FamilySearch and the FamilySearch tree that you mentioned. But before we do that, can you just tell us more about some of these projects? Like let’s start with the African-American families project, what kind of records are you using for that and how does it work?

Joe Price (11m 15s):
Yeah. So what, what started there is I do a lot of family history activities with youth groups, and I had a great experience to do one with a BYU student stake. So I think it was maybe 800 young adults in the room. And we did the relatives around me on the FamilySearch, and it was such a cool experience. Everyone was just discovering cousins and having a bunch of fun. And then some African-American students came up to me and basically they’re upset. I’m going to say at worse than it was, it was like, you have no relatives in this room. And it was just that the African-American tree on the FamilySearch it’s not as complete, coverage is smaller. So we started to quantify this and it turned out that if you look at the 1900 census, about 72% of white people in the 1900 census had profiles on the family tree compared to only 4% of African-Americans

Nicole (11m 60s):
wow,

Joe Price (12m 1s):
this is a massive disparity. And so we started thinking as a lab, what we could do to help, because what this meant is if I was helping an African-American find their grandma or grandpa on the family tree, they’re going to have a very different experience than if I was helping someone who was white. And so over the summer, this was a blessing of COVID. We had access to a lot of computer labs. We had an access to family history center, and we wrote a program that helped us add families from census records to the family tree. And so we added about 600,000 African-Americans to the family tree. And it’s been really neat to see people crushing into those and ask us questions. Like, are we cousins? But I think it’s created a discovery experience for a lot of African Americans that wasn’t possible before.

Nicole (12m 39s):
That is fantastic.

Diana (12m 40s):
Can I ask a question? When you say you created these families, were you taking, for instance, the 1870 census, the first census after emancipation or the 1880 or are taking all the different ones? How did that work.

Joe Price (12m 53s):
Yeah, I mean, 1870 would have been fantastic. The hard thing with 1870 is it doesn’t list the relationships between people. You can kind of guess it a lot of the time. So we chose the 1900 census for two reasons. I mean, the main reason was it has birth month in there. And so that’s a fantastic thing because it makes it more linkable to see a birth record or a death record. So yeah, we, we basically took a nuclear families from the 1900 census. We looked for families where there was a mom and a dad and at least one child. And those are the families that we put on the tree. Someone was like, well, what about single parents? And we’re like, yeah, you’re right. They should be on the tree too. So we’ve been working on single parents and then someone’s like, well, what about someone living by themselves? And so even though we started with nuclear families, we’ve been kind of branching out from there to be more inclusive.

Diana (13m 36s):
That’s wise, 1900 does give you so much information. And by then a lot of times an African-American family has settled on a cohesive surname because those surnames change sometimes between 1870 and 1900. And so maybe that would be the surname that they follow going forward into the 20th century. So very interesting to know how that came about. And then of course, as that goes onto the FamilySearch family tree, then I’m guessing other record hints start coming up for that family that can be attached and flush out the family.

Joe Price (14m 9s):
Yeah. That’s why we focus on families because it gives you multiple people that can have record hands. And so it’s really great because once you put the whole family on the tree, then maybe one of the kids will, a record hit will show up for their marriage record. And in that marriage record, it’ll list the maiden name of their mother. And so then you can go and put in her maiden name, which is great. Cause then maybe that’ll allow you to find her in the 1880 census. And so really these islands that we put on the tree, even though it feels like it’s unconnected to begin with, it’s amazing how quickly those islands connect with each other and then start to connect with the tree.

Nicole (14m 43s):
That’s what I was wondering is how you add them to the tree. So you add them as an unlinked family and then hopefully are able to link them in somehow or another user could come in and find them and link them to their family.

Joe Price (14m 55s):
Yeah. So we add them as an unlinked family, you know, you’d see a father and mother and then the kids, and then they’d be attached to the census. And then what happens is as those hints start to appear, we share those with some volunteers that are helping us. We have kind of a map app and then a Google spreadsheet. And they’re just basically working to those hints and using those hints to connect the island to other stuff.

Nicole (15m 16s):
Great. So we probably have listeners who are wondering how they can be part of this and how they can volunteer. So maybe in the show notes, you can give us a link and we can put some information for how to have volunteer.

Joe Price (15m 27s):
For sure.

Nicole (15m 27s):
Okay. Tell us more about some of your other projects. Don’t you have one that’s just kind of general about censuses?

Joe Price (15m 34s):
Yeah. So the census project is linking people across all the U S census that occurred during their lifetime. We call this a census tree. And so the way you can think about this is the family tree is a tree that’s been created by people, connecting people together. The census tree is a way of just connecting all the records together. And then for each person, you can see each of the censuses that they show up in, and then you can see who their parents are, their spouse, siblings, and children, and using just these two pieces of information, the link to the census and linked to family members. You can actually recreate family trees. And we think this will be helpful because we want to create a museum experience that when you go to the museum, we would ask you and say like, Nicole, tell me about your grandpa.

Joe Price (16m 14s):
And you’d say, my grandpa was so-and-so and then would basically find your grandpa in the census tree. That way we could check to see if your grandpa’s on the family tree, if your grandpa’s on the family tree then we could show you how you related to people in that museum. Now, if your grandpa’s not on the family tree, what we would do is use the census tree to find the quickest way to connect to someone on the tree and then give you that same experience. So if you think about something like relatives around me, it’s a pretty neat experience, but it relies on you personally, having yourself connected to the tree and the census street could help bridge the gap for people that have maybe never used FamilySearch before, or don’t have a very complete tree.

Nicole (16m 48s):
Neat. So what is the output of this museum experience going to be, as it going to be a website? Where will this census tree be? Is it something you will have only, and then people will come to it through you?

Joe Price (16m 59s):
Our goal is actually that the census tree and the family tree will just slowly get to be more and more similar to each other. So I think our ultimate goal is that the family tree will be complete. We always think of a census tree as like an intermediate step to help people who have never used FamilySearch before. So really I think the magic will happen behind the scenes from the user’s experience, all they’ll know is that they told a computer something about their grandpa and suddenly they were given a list of people in the museum that they’re related to. And how, and so we think this will make museums a more personal experience and might draw younger. People do use the museum when they see that they’re part of these things and maybe make history more real for them

Diana (17m 40s):
Right now. That is just a project that’s living on your servers there at Brigham Young University. That’s not anywhere public yet.

Joe Price (17m 48s):
Yeah. It’s, it’s not living anywhere public, but it’s informing the way we find record hints. And so when volunteers help us do record hands, really the census projects happening in the background. And now we’re trying to find ways to make it a more public experience. It’s a little harder with a lot of the museums closed, but like our hope is once, you know, life goes back to normal, we’ll be able to start to use it in that way.

Diana (18m 9s):
So you are hoping to partner with museums all across the country and that when people came in, they would have this experience.

Joe Price (18m 16s):
Yeah. The hope would be to be print out. They’re there four Gen chart for them, maybe create a list of people in the museum that related to. We’ve done this with the BYU art museum. So we’ve been kind of piloting it there and it’s a pretty cool experience. So yeah, I mean, this is a case where I guess if a listener was like, Hey, we’ve got this really cool museum in our town. They were willing to reach out to the museum. We would love to help make something possible.

Diana (18m 37s):
That is so interesting. And if it was just a matter of having a computer and being connected to the website, it sounds like it would be pretty simple for even just a small little historical museum.

Joe Price (18m 47s):
Yeah, for sure. And we even make it, so it works on your phone. So like at the art museum, we have these little QR codes, you just scan them and that shows you how you’re related to the artist. Local museums are actually perfect for this because the people in the community are likely going to be related to the people that picked it in the museum. So that’s going to be a really great experience.

Diana (19m 4s):
Oh my goodness. That’s so fun. Well, I’m thinking of my hometown. I grew up in Burley, Idaho, and they have this little Cassia county historical museum of all these artifacts and my grandpa donated a cabin. And how fun if somebody could go in there and then see how they connect to these people, that their stuff is everywhere in the museum. I can just really envision this being amazing. So thank you for explaining how that works and coming up with this idea, how fun.

Nicole (19m 33s):
All right. Tell us about one more project. What is the Automated Indexing project that you do?

Joe Price (19m 39s):
Here’s the dream for it, think about your attic or your basement for a second, or just those boxes in your closet, and think about all those pieces of paper that you have. Some might be records, some might just be written notes or things. What if we could scan all of those and convert them into searchable data. Now that’s still pretty far into the future. So we’re trying to move in that direction by thinking about large data collections that we can automatically index. So right now we’re doing the 1940 census, which has been indexed, but there’s a whole bunch of rich fields in the index, like occupation, income industry, and such. And so we’re teaching a computer how to find those fields and extract the information.

Joe Price (20m 19s):
We’re also using the same tool to do countries where we have not indexed the census records. So the Czech Republic is a country that would be so neat to provide a discovery experience for people, but the indexing is going pretty slow. And so we’re teaching the machine how to figure out the structure of the form, extract the name, extract the birth place, extract the birth year and all those other pieces of information. So that’s the machine part of it. And the machine does a pretty good job. So the question is how could a human help the machine? Well, we’ve created a fun way to do indexing. We call it reverse indexing and what we do, like if you were to type in Dyer, then I could show you all of the images that the computer thinks is D Y E R. And then you could look at them and see if you agree.

Joe Price (21m 2s):
And if you disagree, you just click on them and it hits submit. And the act of clicking on them is basically telling the machine you were wrong. And that’s what we call it. Reverse indexing is instead of trying to figure out what it is, you’re just telling the machine what doesn’t belong. And the thing I’ve loved about reverse index scene is that my six year old daughter, who I don’t think anyone would have ever thought as being an indexer, can actually type in PRICE. And she’s gotten pretty good to know what Price looks like. The other cool thing about this is it works really well on a phone or a tablet. And so now it’s being put on tablets that are going into the prison system. And so this is a form of service that those in custody can perform. And I think this is a population that, you know, in the past for them to do indexing, they’d have to go into a computer lab.

Joe Price (21m 45s):
And right now those labs are kind of locked down because of COVID. But with the tablets, we might be able to reach more people, give them a small way to contribute.

Nicole (21m 54s):
Wow, you said so much I want to respond to. The first thing is that I have tried to reverse indexing because I saw a link on Twitter and it was kind of fun. It’s like a game.

Joe Price (22m 6s):
It is like that. We do have a, we’re creating a VR version of it. So if you want it to be even more like a game, the VR version, will serve as that,

Nicole (22m 15s):
And the other thing that I love is that you’re providing this to the inmates in custody that they can work on it too. I think it’s really neat that they can have the opportunity to, to help even when the computer labs are closed.

Joe Price (22m 28s):
Yeah. We’re really excited because there’s so many tablets in the prison system right now, and this, the prison system needs a way to provide meaningful service to the inmates so they can maybe earn their time on the tablet to do other things. So I think this is just a huge win-win, but ultimately it’s like, how do we involve more people in family history so that their hearts can be touched and I’ve seen the impact of indexing on a lot of people. And so I’m just trying to think of how to make index kind of an easier entry point because reverse indexing there’s really no instructions. You just kind of open it up. You don’t really log in, you know, and not everyone likes that approach. But what I like about it is it allows me to work with a large group and just have them try it out for a few minutes.

Nicole (23m 7s):
Yeah.

Diana (23m 8s):
So that might be a way for them to go, oh, indexing is fun. You know, they probably heard about it or not really understood what it was and maybe didn’t want to have anything to do with it, but just trying it, then they could see it’s great. Well, I just wanted to comment on the idea of working with the inmates in the prison. I gave a presentation once and talked to a woman afterwards and she had just recently been released from being in prison, and in prison she learned to love family history and it had really changed her life. And it was so neat to hear her story and see how that touched her. So we’d like to talk a lot about family history, healing hearts, and healing families.

Diana (23m 49s):
And how neat, if you know, this could give more people access to that power.

Joe Price (23m 55s):
Yeah, I definitely agree. I mean, the prisons is such a neat place for us to serve people. One experience recently that FamilySearch gave me is they shared 20 letters with me, that people in custody had sent them, you know, asking questions about their family. And it was a really neat experience to do some of that, what I would call like basic family history discovery work, helping them, you know, know who their parents or grandparents were. And so we’ve been trying to think of ways to involve more people in helping people in custody in that way.

Diana (24m 26s):
That’s wonderful. Well, let’s get to your latest project, which might be connected to our current situation. And that is the families of the 1918 pandemic.

Joe Price (24m 36s):
Yeah. That that’s been a really neat project to work on. And it came about just by chance. Well, if you think about the things that were in place, it’s kind of amazing. We, we already had an NIH grant to figure out how to automatically extract cause of death from death certificates. So FamilySearch had already indexed death certificates, but they’d never indexed the cause of death. And the cause of death is actually really helpful for a lot of public health projects. And so we were already working on this. This was part of a project with Yvonne Roberts at Minnesota and Martha Bailey at Michigan. So we already had kind of some of the tools in place and then the pandemic hit and we thought, oh, maybe this data could be used to study the 1918 pandemic.

Joe Price (25m 20s):
And so we extracted all that data for Ohio. And then we reached out to FamilySearch to see if we could get records for other states.

Diana (25m 27s):
So what kind of things are you finding from this? I’m just curious about any connections that you’re making with areas and localities and the number of people who died. Is this an economic type thing that you’re looking at?

Joe Price (25m 43s):
Yeah. I mean, here’s the nice thing is, let me talk about the advantages of extracting it from the individual cause of death. Cause we already had data at the state level and for big cities for each month during the pandemic. So we already knew a lot about the age distribution we knew kind of about the temporal patterns. What I think is unique about getting at the individual death certificates is two things. One is we can now do everything at the daily level. And so when you go to an evaluate a public health intervention, that can be really helpful to know like here’s one example in one city, the death rates started to shoot way up. And so then they implemented an intervention right in the middle of the month, which brought the death rate down.

Joe Price (26m 27s):
If all you had was monthly data, it would look like in the month that they did the intervention, the death rate was really high. But the problem there is that it was the intervention that was responding to the death rate. And so the monthly data can maybe be a little misleading. So with the daily data, we’re able to really narrow in on the time patterns, right around these interventions. And then the other neat thing that’s great about the individual level data is you can link the death certificate to the 1910 census and then link forward to the 1920 census. And so you can actually learn things about those who didn’t die, who were from the same family, because we don’t know what the consequences were for people who were in the household that lost someone during the pandemic. And then the other nice thing about this is we can link it to the family tree so people can know which of their relatives died during the pandemic or were affected by it.

Joe Price (27m 13s):
And this can open up maybe new discovery experiences as people go back and look at journals,

Diana (27m 19s):
Right. I just read a story this morning about a woman whose husband was dying. They were a young couple, you know, because the type of flu it was in 1918, really attacked the younger people. And she stayed up all night, putting mustard plaster on his chest. And in the morning he woke up and was better. It was just such an interesting story. And I just thought, oh my goodness, can you imagine? And that happened to our ancestors all the time, where they didn’t have a hospital to take their loved ones and they just had to provide the medical care. So when you look at those death certificates, sometimes as family historians, we just are so excited to find the date of death and something about birth.

Diana (28m 2s):
But it’s always interesting to think about the actual situation. And some of those are really sad and this pandemic was really, really tough. So, wow. This sounds like an amazing project.

Joe Price (28m 14s):
Yeah. I mean, I’ve had my own heart touched because again, once we found the people in the death certificate that died from influenza pneumonia in 1918, we would then create a profile for them on FamilySearch. And then once we created that profile, we try to link that person to other records, maybe a birth certificate or their parents in the census. And so as a result, we ended up kind of reconstituting families. And I remember I was working on one family and they’d had three children die during that fall of 1918. And I just remember like, as a parent, like having my heart break in the moment and I think it just opened up all the empathy for what people go through during a pandemic. So even though the current pandemic hasn’t affected my family very much, it definitely opened my heart to be more willing to help others who maybe are going through something really hard because I just felt like it just felt real when I looked at those records.

Diana (29m 7s):
Yeah. How would one of our listeners know to look at this family of the 1918 pandemic project? Is there something online we can look at to see what’s going on with it?

Joe Price (29m 16s):
So you can go to the website. Now the, the thing about the website is you kind of have to know where your ancestors were living during the pandemic, because right now you search by state and county. And then once you get into the county, you can see the list of all the people. I think our hope is in the future to make it more searchable and also maybe embedded within relative finders. So people can see how they’re related to these people. The problem is there were so many people that died that the computation like really to find it works great for small groups like artists and politicians. It gets a little computationally expensive to do it for a big group. So I think the easiest way is for people to go, maybe look for the people with their family name in a place where they know their family’s from, or I think the other way is we’re all part of a big family.

Joe Price (29m 57s):
So maybe go look at the place where you’re from. You touched one of those states, you know, there’s 12 different states and you can go and look and see who in your community died during the pandemic. And the neat thing is you could even go to the family tree for that person and help grow out their tree a little bit, which will make it easier for other people to discover their connection.

Diana (30m 16s):
Okay. So we’re going to link to this website on Family Locket so you guys listening can all go look for your people, but I’m just looking right now at my home county Cassia county in Idaho. And my mom tells me all the time that she was too young to remember it, but she remembers people talking about the boy down the street who died. And I’m seeing that there are 38 records for Cassia county in Idaho, people that died in the pandemic. So that’s so fun to be able to go to your home place. And I guess I’m lucky that Idaho’s one of the places that you were able to index.

Joe Price (30m 49s):
Yeah. And now one thing is if your state’s not on there and we basically gathered all the records that we can get public access to. So if your state’s not on there, you might ask your state archive, hey, can we be part of this thing at BYU? Because if any state archive were to reach out to us and just say, hey, here’s the images for 1918, we would process them for sure and put them on the website.

Diana (31m 9s):
Oh, that is great to know.

Nicole (31m 11s):
Can you tell us more about the technology that’s working with FamilySearch to transcribe these death records?

Joe Price (31m 17s):
Yeah, it has kind of a three-step process. So the first step is the skewing. So basically the images aren’t always perfectly straight, so I don’t need to straighten them out a little bit. The second part is segmenting, which is figuring out exactly where that cause of death is on the record. And so what we’ll do is we’ll have students label it by hand, you know, like a couple of hundred of them, and then the machine can learn from that to know where to look. And then the last part is the handwriting recognition. So once we know where to look, we cut a little box out of the image, and then we read the words in that little snipped image and it’s never perfect, but the nice thing is for influenza and pneumonia, they’re such unique combinations of letters that we were able to actually flag that with almost perfect precision.

Nicole (32m 0s):
Wow.

Joe Price (32m 1s):
And now we’re trying to apply that same approach to other things like stillborn or, you know, other common diseases, but ultimately, and the machine just gets better and better over time. So the more examples you show it, the better it gets at reading the handwriting.

Nicole (32m 16s):
It’s so fascinating to think about machine learning and that a machine can get better.

Joe Price (32m 21s):
Yeah. Never forget, machines are like pretty amazing, I mean, they’re kind of dumb in a lot of ways, but they’re pretty amazing in one way, which is once they’ve learned something, unless you tell it to forget it, it can hang on to that information forever.

Nicole (32m 34s):
That’d be great.

Diana (32m 37s):
Yeah. I could really use that skill. Well, just tell us a little bit as we’re kind of coming to the end of this podcasts about how this project can really benefit family historians, you know, this 1918 pandemic project,

Joe Price (32m 50s):
Here’s the bigger picture that I would say. We’ve always had people that have been interested in family history because they love family history. I think there’s a whole other group of people that we can bring in to this experience by giving them something that doesn’t feel like family history, right. In the beginning, imagine I approached a bunch of public health people and I say, hey, we could learn a lot about how to deal with pandemics by using this data. This data happens to come from family history. And so suddenly now we’ve taken family history, which has one purpose and given it another purpose. And I think the more and more purposes that we give family history, the more we’re going to reach different groups that we’ve never reached before. So I love giving academic talks because I’m often speaking to an audience that has never heard of family search or never done family history.

Joe Price (33m 33s):
And I show them what we can do as economist with family history. But my hope is always that your heart will be touched as you’re thinking about that use of family history. And it might draw you back to using family history for more personal things. I also think like a project like this highlights what’s possible, we literally had a machine read images and extract a piece of information. And it was a very narrow scope that we had, but it brings hope that we could then use the same approach on other records and continue to expand what the machine can do in terms of auto indexing.

Diana (34m 6s):
Yeah. I remember hearing David Rencher talk about this auto indexing a few years ago and I’m thinking, wow, that would be so neat. And it’s exciting to hear that you are actually putting it into practice and I’m sure you are learning so much with every project, how to make it more effective and better. So, wow. Thanks for sharing so much about that with us today.

Joe Price (34m 29s):
And I guess one thing with that, Diana is I really think the auto indexing is going to move faster than we ever thought possible and touch us in ways that we wouldn’t have imagined. So here’s one thing I gave to my dad for his birthday this last year, he lent me his mission journals. And so I took pictures of all the pages in his mission journals. And then I taught a machine to read his handwriting and it took a lot of training. You know, I had to kind of teach the machine a lot, but in the end, the machine was actually pretty good at reading his journal. And so I was able to give my dad back a fully searchable journal. And so he can go and look up specific experiences and places and people. And then I was able to start to use the people in that journal and try to look for them, either them or their, their descendants. And that’s been a really neat experience to think about my dad’s journal.

Joe Price (35m 13s):
It’s just one little record that nobody has ever had access to. And it mentions a whole bunch of people. And so people can learn about themselves by looking at that one record. And so I think historically what we’ve done is family historian says, go try to look for the needles and the haystacks, but there’s a gajillion haystacks to look in. I think what would be really neat is if all of us just grabbed the haystack that was closest to us and grabbed all the needles in that haystack and stuck them on the family tree or made them searchable so that everyone has access to every needle everywhere in the planet. And I actually think that way of doing family history might be closer than we could imagine.

Diana (35m 48s):
Well, that is super exciting because we talk a lot about going to archives and looking for manuscript collections, things like your dad’s journal that have the clues to our ancestors way far back. And so, wow. How exciting would that be if in all these different, like you said, we all have our own family archives, right. But then as that expands to archives and libraries and museums around the world and more of those materials become available, maybe we will actually be able to solve more of our family history mysteries that we’ve been puzzled with for so many years. So exciting.

Joe Price (36m 25s):
Yeah. And I think it’s also a compliment to professional genealogists, which is it’s going to draw more people into the family history umbrella, and as people come in, they’re going to have harder and harder problems to solve. And so I actually think that there’s going to be a synergy. That’s going to be a win-win for everyone.

Diana (36m 41s):
Yeah. And I do see that in our professional work that a lot of clients have the actual materials, but there’s just something with a genealogy brain that’s able to put all the pieces together. And so someone might have all the pieces, but not know quite how they all fit. And so I think there’s probably going to be a place for everybody.

Nicole (37m 1s):
I love the idea of grabbing whatever’s closest to us and making it available on FamilySearch. I’ve just been thinking about that the last couple of minutes here. And you know, one of the things that Diana and I have talked about in the past is what to do when we come across records of enslaved people in our own family trees, because we have ancestors from the south. And so we’ve always kind of wondered like where is a place where we can put the names of these people. And there used to be a blog that would have a way for you to submit names of enslaved African-Americans. And I don’t know if that’s active anymore, but you know what the FamilySearch tree really is the perfect place to add them.

Nicole (37m 41s):
Just like you’ve been doing with some of these other projects where they’re unlinked at first, and then hopefully somebody comes in searching for their family and finds them and can link them up.

Joe Price (37m 52s):
Yeah. I mean, wouldn’t it be neat if that was the way everyone did academic research in the future? Like if you were studying a particular insulate person, would it be neat if you could just go on the family tree and have access to all of those different documents about their life? I kind of envisioned like David McCullough in the future, like, you know, he’s about to write a paragraph about someone. And so he goes on FamilySearch really quick and he can add access to those things. I mean, that’s my dream. Would that be neat if the academic community thought of the family tree with as high regard as they do other types of archives?

Nicole (38m 21s):
Wow, that’s a fun take on it.

Diana (38m 23s):
That’s good. That is awesome. I love that idea that FamilySearch can become this great archive and, you know, because it is collaborative and everyone can add their information and their ancestor. And it really speaks to just education teaching people, how to use it and how to analyze and make sure the record that they are touching is really for the person there’s some education there, which is what we’re all about. So it’s great to see the types of things you’re doing, come together with all these other projects out there in the world. So yay.

Nicole (38m 57s):
So Joe, what future projects can we expect from the record linking lab?

Joe Price (39m 2s):
Yeah. I’ll mention three. One’s small, which is, I think we should start to think about groups that will be blessed by family history. So two groups that I’ve been thinking carefully about these people that are incarcerated and people that are homeless. And I guess what I encourage people to do is think about what are the groups that you care about and how can you help them. So we recently reached out to homeless shelter and we provided a way for people just to tell us their story, like tell us about what you know about your family. And then we’re showing them their family tree. So that’s what I would call like a very personalized experience that we want to create the tools so that people anywhere could provide that service to people in their community. I think if we go one more step up the other projects, we’re doing our community reconstitution projects.

Joe Price (39m 42s):
So this is again using machine learning, but also human effort to try to make sure that everyone living in a particular places on the family tree. So my son’s living in Skagit county, Washington right now for a couple of years. And so I’ve been helping add every single person that lifted Skagit county to the family tree from the censuses and other records. And now we’re bringing in high school yearbooks and obituaries and biographies. And so now anyone that has family from Skagit county is going to be able to find their family. So I think where we see our lab going is creating the tool so that anyone anywhere could do the county, they care about or love and make it a discovery experience for others. And then the third level is what can we do this for the whole world?

Joe Price (40m 25s):
And the question is, is it possible to gather every person that’s ever lived on earth into one place? And the place where we’re going to start is something called the database of human lifespan. And so we’re, we’re now starting to pull together as many death records as we can around the world and see if we can gather usable data about human lifespan, which we think will attract the interest of scientists, but we’ll ultimately gather together the data needed to build the family tree for the whole world.

Nicole (40m 52s):
Well, those are some really amazing goals. The one I want to mention real quick is when you mentioned the homeless people, and we actually had a really interesting guest blog post in December about a group that’s doing that in Sacramento called Olive Branch Connections. And maybe Diana could tell us more about what they do.

Diana (41m 18s):
So this non-profit does research for people in the homeless shelter and when they’re ready to leave the shelter, they present them with their family history and give them stories. And they’re looking specifically for stories about ancestors that overcame really difficult things, because that gives a person strength that they too can overcome. And so anyone who wants to participate in doing research and trying to find the family tree of a homeless individual can be part of Olive Branch Connections. So the article just really discusses the great blessing. It has been to, to a person who temporarily, or maybe habitually is homeless and is filling, you know, hopeless and, and full of despair and how it has really changed their lives.

Diana (42m 9s):
So that’s neat that you have also been thinking along those same lines of working with the homeless population and how family history can help to bus their lives.

Joe Price (42m 17s):
I mean, those are the groups that we love hearing about. Cause, cause I think when you hear about a success like that, then it encourages other people to do the same. And I guess the way I view the future of my lab is to be kind of maybe the little magic that happens in the background that you don’t really notice, but it maybe helps accelerate the work that groups are doing.

Nicole (42m 36s):
Thank you so much, Joe, we have loved hearing about the magic your lab is doing, and it fills me with excitement for the future, with all the machine learning and the wonderful building of the family tree that you’re doing at family search. So thank you.

Joe Price (42m 49s):
Thanks. It was great to be on your show.

Diana (42m 53s):
Well, we’re excited to hear more and maybe in a year or two, we’ll have you back on, and you can tell us about your next latest and greatest project because it sounds like the ideas just keep coming to you.

Joe Price (43m 3s):
Or to the students. I mean, the nice thing is I’m around some very young, young, energetic people that don’t know what’s possible or not. They do dumb things, but they sometimes will do things that blow my mind. And so

Nicole (43m 13s):
That’s great. I love it. Well, thank you so much and to all our listeners have a great week and we’ll talk to you again next week.

Diana (43m 22s):
Bye bye everyone.

Nicole (43m 23s):
Bye bye. Thank you for listening. We hope that something you heard today will help you make progress in your research. If you want to learn more, purchase our book Research Like a Pro a Genealogist Guide on Amazon.com and other booksellers. You can also register for our Research Like a Pro online course or join our next study group. Learn more at FamilyLocket.com. To share your progress and ask questions join our private Facebook group by sending us your book receipt or joining our eCourse or study group. If you like what you heard and would like to support this podcast, please subscribe, rate, and review. We hope you’ll start now to Research Like a Pro.