Family Historian User Group

There's an interesting series of articles / posts on the blog The Ancestry Insider whereby the AI talks about the handling of evidence. This is inspired in part by Elizabeth Shown Mills' 2009 book Evidence Explained: Citing History Sources from Artifacts to Cyberspace.

This link contains the Table of Contents with links to all the posts in the series so far:

http://ancestryinsider.blogspot.com/201 ... ement.html

For a concrete example of Evidence Management, see this entry:

http://ancestryinsider.blogspot.com/201 ... ained.html

He gives a table for an individual listing all the pieces of information used to determine the individual's birth date. There is a space to explain conflicting evidence and to note how the researcher came to a conclusion.

I've struggled with this ever since I started doing family history, because IMHO all the programs I've tried force the user into making conclusions ahead of the evidence. Some programs allow you to enter multiple entries for 'facts' and choose one as the preferred entry, but what if (as in the example the AI gives) the birth date gets the day and month from one source and the year from somewhere else?

Users could create an 'event' with their conclusion and put the evidence management in a note. But just like research logging, this is a place where computers could help us, but no one has (yet) made the tools.

ID:4642

ooh! A good set of posts to look at -- thank you.

'I've struggled with this ever since I started doing family history, because IMHO all the programs I've tried force the user into making conclusions ahead of the evidence.'

That's because they all use Gedcom as their initial starting point. Gedcom is not an evidence model, it's conclusional. The other thing that affects how family history* programs are written is that the target market is those just doing their immediate family history. They want to see immediate results and forcing them to put in the evidence to back up all the stuff that they know from personal recollection is going to put them off.

* I say family history to differentiate from genealogy. One is a simple hobby, the other is a lifetime of research! A lot of familt historians catch the bug and become genealogists.

It's an interesting series of articles on that blog. I don't know if I'm being pragmatic, realistic or defeatist but I suspect little will come of these ideas for precisely the reason Jon alludes to. All(?) FH programs are based round GEDCOM. This generates a fairly tough series of requirements on any software incorporating Evidence Management.

1) It must be able to import GEDCOM files, break the current links apart and insert some evidence summaries / assertions / whatever, concocted out of the current file. Because if I need to re-enter my data from scratch - I won't.

2) It ought to be able to export GEDCOM 5.x files, so the evidence managers can send stuff to others and use stuff like TNG.

3) It ought to have the means to export and import evidence management ('EM') type files between itself and other 'EM' type software. This implies an equivalent to GEDCOM to incorporate 'EM'. (Who'll create that?)

4) It must have the ability to be used in a simple fashion by those who haven't much grasp of Evidence Management, while storing the data in such a way that when they do understand it, they can drill down and expose the assertions / assumptions / whatever that they made but never realised, then work in a fully 'EM' compliant way. (Why? Simply because no-one will ever get a cash return in selling software to the master logicians only - if they could, then I presume that TMG would have a more up-to-date interface than it does! And I think we all start like Jon's example, just putting in our family details.)

Years ago, there was a GENTECH produced assertion-based data model rather similar to this. A model is all it still is.

(Incidentally Jon - Curious you have FH and Genealogy that way round. I've always thought Family History as a more sophisticated version of Genealogy - it's not just who begat who but the full context. The American Genealogists with their Chicago Style Manual citations leave me cold - they miss the point with rules like 'italicise the title if it's unpublished'. And you can have perfect citations for garbage logic!)

Adrian Bruce said:
(Incidentally Jon - Curious you have FH and Genealogy that way round. I've always thought Family History as a more sophisticated version of Genealogy - it's not just who begat who but the full context.

I use the term 'family history' because as far as I am concerned, I am studying history -- with the focus on the family unit (as opposed to a larger unit). I employ what some call 'cluster genealogy' as a means of doing so -- studying not just a single family unit, but their neighbors, friends, business associates, etc.

I suspect this is not what most people have in mind when they use the term.

I forgot to say before that this kind of evidence management is not (as far as I can tell) new to genealogy. I've seen these techniques used in paper-based methodologies (e.g. the book Unpuzzling Your Past).

What baffles me is that no one (apparently) has translated these tools to the computer world -- not even as far as translating the worksheets in Unpuzzling Your Past to Excel templates. I've seen Excel spreadsheets for transcribing things like the Census, but not for research logging or aids for analyzing evidence -- at least not with the ease of use of FH or GC/AS. [wink]

While searching for the name of a program I tried once, I discovered that Louis Kessler (author of Behold! an interesting program which takes a GEDCOM and turns it into something readable by a human) has completely redone his genealogy software review site. So perhaps it is time to take a fresh look at what else is out there.

See, I want something like THIS that lives on my own computer and not in the 'cloud' -- although online sites like the family trees on Ancestry.com SHOULD have an apparatus like this. Check out this demo of Lineascape by Ben Sayer:

http://www.lineascope.com/help

Don't get me wrong -- I like Family Historian, otherwise I wouldn't have chosen it over all the other programs I have tried. But if you've ever wondered why on earth some of us use multiple birth events and so on, it's because we're trying to do evidence management.

Thanks for that Lineascope link. I think it interesting that (if I understood it correctly) the author was not envisaging it being used for every 'fact' in a 'database', rather as supporting explanations for some major elements.

That could help with the workload because there's no doubt that doing Evidence Management 'properly' is going to be more work. (And yes, conversely, you justify it to yourself by saying that you save time in the long run by not making mistakes). If you only do this for major things like 'proving' someone's parents, spouses, etc, and don't do it for their occupation / address / etc in a census year, you save time.

It also steps round the GEDCOM aspects because the 'proof' just becomes a free-standing multimedia item linked to the GEDCOM.

So it's an interesting approach - except what worries me is whether this approach simply makes Evidence Management a spray-on extra instead of an integral part of things. But since I remain to be convinced that EM can become an integral part of things....

Yes - I'm firmly still sitting on the fence!

To me it makes explicit the process that takes place (or should be taking place) before one puts information into the conclusional program.

(Note that for FH, we have auto-source citation, whereas for other programs citing sources properly is a spray-on extra. Evidence management could be treated as a spray-on extra, but it needn't be so.)

jmurphy, thank you for passing on the link to Lineascope.

Would you mind telling me why you want something like Lineascope, except on your computer? I assure you I'm not trying to defend the choice—I'm actually not committed to it. I want to understand your perspective so I can make better decisions in the future.

jmurphy said:
See, I want something like THIS that lives on my own computer and not in the 'cloud' -- although online sites like the family trees on Ancestry.com SHOULD have an apparatus like this.

Adrian Bruce said:
Thanks for that Lineascope link. I think it interesting that (if I understood it correctly) the author was not envisaging it being used for every 'fact' in a 'database', rather as supporting explanations for some major elements.

Adrian, I envision Lineascope being used as a research tool for any 'fact' one cares to prove. I tend to agree with you that some events, characteristics, and relationships are more important than others, so I leave it up to the researcher to decide when to use the tool.

Adrian Bruce said:
It also steps round the GEDCOM aspects because the 'proof' just becomes a free-standing multimedia item linked to the GEDCOM.

Exactly. As was pointed out earlier in the thread, GEDCOM is based on a conclusional data model. Lineascope is aligned with the GENTECH data model that you previously mentioned.

Adrian Bruce said:
So it's an interesting approach - except what worries me is whether this approach simply makes Evidence Management a spray-on extra instead of an integral part of things.

I would like every genealogist and family historian to make evidence management an explicit part of their research method. I also recognize that people are meeting different needs when they conduct research, so that's not going to happen. So I choose to help the researchers who do practice EM or want to, but don't know how.

I've observed a lack of tools that hampers those people. So I am creating Lineascope to be a general purpose EM tool. To be sure, it's a first attempt, so all the feedback I can get will help.

How can we make evidence management an integral part of things?

Adrian Bruce said:
This generates a fairly tough series of requirements on any software incorporating Evidence Management.
1) It must be able to import GEDCOM files, break the current links apart and insert some evidence summaries / assertions / whatever, concocted out of the current file. Because if I need to re-enter my data from scratch - I won't.

I've thought a lot about this. I think importing GEDCOM data does not provide the information necessary; evidence can't be correctly inferred from sources and conclusions. That's why I created Lineascope—to capture what is otherwise often missed.
I feel your pain regarding re-entering your data. Given that

• the data regarding the evidence isn't in the GEDCOM export
• one wants to conduct research about an event, characteristic, or relationship for which some research has been done
• duplicate entry should be minimized

How could one go about using an evidence management tool to do the new research?

How about loading the GEDCOM, identifying the subject and event, characteristic, or relationship which is the relevant to the research question, and importing only those pieces of information. For example, when researching a birth date and place, import the GEDCOM records for the individual, the event(s) (and respective sources and notes).

Adrian Bruce said:
3) It ought to have the means to export and import evidence management ('EM') type files between itself and other 'EM' type software. This implies an equivalent to GEDCOM to incorporate 'EM'. (Who'll create that?)

I'll do it.

Adrian Bruce said:
4) It must have the ability to be used in a simple fashion by those who haven't much grasp of Evidence Management, while storing the data in such a way that when they do understand it, they can drill down and expose the assertions / assumptions / whatever that they made but never realised, then work in a fully 'EM' compliant way.

What's missing from Lineascope in this regard?

Addressing the question of cloud vs. non-cloud, I'd rather use:

• something I can carry with me everywhere not just where there's cloud-connectivity.
• something where I control what happens to the data -- particularly privacy and security including how it's backed up, who has access etc
• something where I can extract the data easily in a format that suits me and manipulate it/report on it as I please

I do like the look of Lineascope but not enough to overcome my preferences above.

Ben Sayer said:
jmurphy, thank you for passing on the link to Lineascope.

Would you mind telling me why you want something like Lineascope, except on your computer? I assure you I'm not trying to defend the choice—I'm actually not committed to it. I want to understand your perspective so I can make better decisions in the future.

Hello, Ben! Welcome to FHUG.

While I am not trained in genealogy, I have an academic background (my degrees are in Classical Studies and in Linguistics). While my instructor in data-gathering did make use of computers as a tool, the research journals and the methods we were taught to use for our analysis were kept entirely on paper, and if I were to bring her approach to genealogy, my primary work would be recorded on paper as well.

Let's say I went to the library and collected an obituary. I might take a notepad with me and transcribe the text, or make a photocopy of a newspaper page or make a printout from microfilm. I would then take the raw notes of what I had collected and copy them into a research journal. [Proper research journals are bound notebooks, with sewn signatures and numbered pages, so that you can't tear out any pages without noticing that something has gone missing.]

Citing one's source is critical, because if you discover later that the person you were interviewing was feeding you mis-information as a joke, or drunk, or otherwise unreliable, you have to be able to toss out all the data that you have collected from that source. (The reason I chose Family Historian over the other lineage-linked software I tested is because of its Auto-Source citation feature -- it allowed me to enter the source first and then extract the date from it, instead of having to work the other way around.)

After tipping in the photocopy of the text, or copying in the transcript I had made, I would then extract the data from the obituary in a process similar to what you did in your Lineascope demo, with my comments about the quality of the data on a facing page.

The chief difference between doing it on paper, while keeping a strict accounting of what you were doing in a research journal, vs. doing it straight into the computer, and keeping your work only on the computer, is that with a research journal you have what a strict accounting program would call an audit trail. You keep a chronological record of what you searched for, what you found, AND what you didn't find. If you have to go back later, you can see not only what you found, but how you got there -- or if you made a wrong turn, how it happened. It's like being able to play back a Google Wave from start to finish, only executed the Old School Way. [wink]

From there we took the information that would be equivalent to the name, date of death, place of death, and so on, and made coded slips of paper, so that for a particular person, we could pull out references to all the sources which had the date of death in them. This is the part of the process where a computer program like Lineascope might be used instead. Since the real work is what I record in my research journal (which would always be in my possession), and the computer is the analog of the paper slip file (which is just a tool used to analyze data extracted from the journal), I suppose it doesn't really matter whether I use a machine which belongs to me vs. someone else's machine. However, I am old-fashioned enough to want to be responsible for all the parts of the process myself, rather than trusting it to someone else.

It is convenient to be able to log into a Google account and get at things, but what if my broadband goes out? If I have no local copy, then I have to go out to the library or somewhere else in order to get at my own work. Thus having it online, and having to go through an intermediary, just feels wrong to me. In the days where I did the analysis on paper slips, I wouldn't have stored those slips in someone else's office -- I would have kept them in my own office or dorm, where I could always have access to them.

(On the other hand, my teacher might disagree with me -- she might say that as long as you had your research journals, you would have control over your real work, and the location of the slip file (or computer equivalent) wouldn't matter.)

At any rate, I prefer keeping local copies whenever possible -- look at what happened with Ancestry, when the Drouin collection was taken down -- all the people whose work was ONLY stored in an Ancestry family tree, and didn't keep local copies, lost the ability to review the material in the documents they had attached to their family trees. Or let's say that because of a DNS screwup somewhere, you can't get through to Google or Ancestry -- you won't have access to your To Do Lists then, if the only copies of them are online.

The recent Sidekick debacle was another incident where people keeping their information in the cloud lost all their data through no fault of their own. Even if the risk that my stuff will be lost through a mishap is greater if I keep it on my own computer, I'd rather be responsible for it myself, because it's my stuff. YMMV.

Jan

Oh, and before I forget, I did want to mention two things about your demo which bugged me. I think it is a mistake not to record negative search results.

Many online collections are constantly being updated, and you may have to go back to them more than once -- if you don't record your negative search, and especially the date that you did the search, you can't look back and see that it has been a while since you searched, and it's time to do it again.

Also, if you get into the habit of making explicit the reasons why your search might have failed, that in itself can lead you to other information (e.g. you didn't find the event in that particular Ancestry collection because the event took place in 1968 and the collection you were looking in begins with records in 1970).

I want to get hack to a point Jon made earlier --

JonAxtell said:The other thing that affects how family history* programs are written is that the target market is those just doing their immediate family history. They want to see immediate results and forcing them to put in the evidence to back up all the stuff that they know from personal recollection is going to put them off.

FH doesn't force you to use Auto-Source Citation -- it is a feature that you can use if you want to. In the same way, EM would come into play most of the time when you have conflicting or incomplete evidence.

Note that in Ben's demo on resolving someone's date of death, he notes that the SSN on a death certificate matches the entry he found from the SSDI. If you are putting in the date of death of an immediate family member, e.g. one of your parents, you already know the date of death from personal experience, and you are likely to have their SSN on documents already in house. You are not as likely to need to construct a formal proof that you've found the correct entry in the SSDI when you are the person who is the informant on the death certificate and notified the Social Security Administration of the death. So I think it's a safe bet you'd skip the evidence management for the date of death in that case.

Ben Sayer said:

Adrian Bruce said:
4) It must have the ability to be used in a simple fashion by those who haven't much grasp of Evidence Management, while storing the data in such a way that when they do understand it, they can drill down and expose the assertions / assumptions / whatever that they made but never realised, then work in a fully 'EM' compliant way.

Ben Sayer said: What's missing from Lineascope in this regard?

To be honest I can't point up specific issues with Lineascope's ease of use by a novice (nor was I intending to) since I've not tried it, I've only thought about it. I was simply putting a requirement on any piece of software attempting EM.

If we acknowledge that there are people out there whose grasp of logic is so flakey that they keep repopulating the LDS's newFamilySearch data with garbage simply because they've seen it elsewhere 'so it must be true', then even normal people are going to need careful training in EM - maybe wizards, etc. I know adepts tend to look down their noses at wizards, but we all have to start somewhere.

The other part of what I wrote above ('storing the data in such a way that when they do understand [EM], they can drill down and expose the assertions / assumptions / whatever that they made but never realised') would neatly be circumvented by the Lineascope approach of separate logic as it would never happen!

Ben - just one thing that maybe I missed in your demos - I wasn't sure if you adequately demonstrated that various bits of evidence belonged to the same guy, not just someone with the same name.

Adrian Bruce said:

Ben - just one thing that maybe I missed in your demos - I wasn't sure if you adequately demonstrated that various bits of evidence belonged to the same guy, not just someone with the same name.

One thing I like to do when I am stuck on a problem is to write it all out as if I were explaining it to someone else, someone who has never looked at any of the data before. Write everything down, make all the assumptions explicit, be as detail-oriented as you can. In the middle of doing that, data you overlooked often reveals itself.

Even experienced researchers can skip a step and get themselves into trouble. IMHO no one demonstrates enough the reasons why they think the evidence belongs to the same person and not some other guy. It's certainly not true of the US version of WDYTYA?

In Ben's demo, the Social Security Number on the death certificate matches he found from the SSDI. I think he could have emphasized a little more that this is something that needed to be checked once he got the copy of the death certificate, because a lot of newbies will simply pull the index records off Ancestry, see that the dates match, and stop there. A death certificate can also give you the place and date of burial, which can be correlated by the entries on Find-A-Grave.

I learned when I was teaching that students often skim and pass over a lot of the material in a lesson, so you really do have to hammer some points home multiple times. But if you have a demo like Ben's which is primarily used to demo the software, you also need to be concerned about the pace of your demo -- if you get too bogged down in details, people may bail out in the middle.

It's a real challenge to do a demo which is sound and doesn't skip steps in the process without making your audience's eyes glaze over.

jmurphy said:
Oh, and before I forget, I did want to mention two things about your demo which bugged me. I think it is a mistake not to record negative search results.

I agree that not recording negative search results is a mistake. What's preventing you from entering negative search results in Lineascope?

Adrian Bruce said:
Ben - just one thing that maybe I missed in your demos - I wasn't sure if you adequately demonstrated that various bits of evidence belonged to the same guy, not just someone with the same name.

You're right; I stayed quite clear of that. [grin] The demo was quite long already and that subject is complicated. I could write a book about just that topic.

I tried to keep Lineascope agnostic to how you assert what evidence relates to what individual. I plan to add the ability to make assertions that group individuals into a single persona as the GENTECH data model specifies--that is if Lineascope proves useful to people. I try to treat everything as a test; I think what Lineascope does is needed, but I could be wrong, so I'm implementing what I think to be the most valuable functions and will see what happens.

The input I'm getting from you and Jan is very helpful. Thanks!

Ben Sayer said:

jmurphy said:
Oh, and before I forget, I did want to mention two things about your demo which bugged me. I think it is a mistake not to record negative search results.


I agree that not recording negative search results is a mistake. What's preventing you from entering negative search results in Lineascope?

Nothing at all. Forgive me if I'm mis-remembering what you said, because I only did watch the demo one time. Clearly in the context of the demo, you must have meant that you weren't going to take the time to demonstrate how one could record negative search results -- but something about the way you said it left me with the impression that you didn't think it was particularly useful to do so.

It has been long enough since I viewed the demo that I can't remember why. Sorry. [frown]

Adrian Bruce said:
Ben - just one thing that maybe I missed in your demos - I wasn't sure if you adequately demonstrated that various bits of evidence belonged to the same guy, not just someone with the same name.

Ben Sayer said:
You're right; I stayed quite clear of that. The demo was quite long already and that subject is complicated. I could write a book about just that topic.

Yes, but shouldn't the issue be addressed somewhere, especially since Lineascope is one of the tools we users might want, in order to solve that very problem?

It is tempting to say 'no one ever talks about this' because many of the authors of the books I've read do address the issue -- obviously, they aren't under the constraints of someone making a demo on the net and have more space in which to do so. But shows like WDYTYA? and other online tree-building sites just sort of wave their hands and gloss over the whole thing, and newbies who are coming to genealogy from the online world often aren't aware of the issue at all.

We've all got our examples of really obviously wrong things on Ancestry, I'm sure -- mine is my husband's great uncle. Some person has gathered up records which I'm 99.99% sure are his and attached them to his profile in their tree, along with a census record from a fellow with a similar name who was born in a completely different part of England. If that person had used Lineascope to work out his place of birth -- if they had actually read the records they attached to this profile -- you would see:

• WWI and WWII Draft Registration cards, both listing him in as having been born in South Devon, although the place in the WWII card is from a different (but nearby) town from the WWI
• US Census Records from 1920 and 1930 showing him living all in the same town in Massachusetts as the draft cards -- matching street addresses make this a pretty sure bet this is the same guy
• a Boston Passenger list for his second arrival to the US, in 1905 which lists both his origin in South Devon and his Massachusetts hometown, which states explicitly that he has been in the US before, from 1898-1904

To this collection of records, this person has added two UK Census records, one from 1891 and one from 1901, to a fellow with the same first name and middle initial who was born in London, and a FreeBMD birth registration from somewhere else entirely.

Whether or not software should force good practice is open to debate. Clearly if the program is too strict, users won't use it -- e.g. GnuCash, which (at the time I last tried it) wouldn't let me create a vendor card without entering the payment address. But I do think that the tree-builders should encourage good practice -- which is why I was tremendously excited to see the discussion on the Ancestry Insider blog, and to see your demo of Lineascope. If my hapless fellow Ancestry user had made a little chart working out the birthplace of her person, or had sat down to make a time line on paper using ALL the information in the records, not just the bits indexed by Ancestry, perhaps it might have become more clear that she is looking at the records from two different people.

So while I understand that for the purposes of your current demos, you didn't want to open this particular can of worms, at some future date, you might want to make a demo showing how Lineascope can help sort out messes like these. It would be a good selling point, IMHO.

jmurphy said:
So while I understand that for the purposes of your current demos, you didn't want to open this particular can of worms, at some future date, you might want to make a demo showing how Lineascope can help sort out messes like these. It would be a good selling point, IMHO.

That's an excellent idea Jan! I have planned on creating a series of videos about doing evidence-driven genealogy with Ancestry.com. I'll be sure that one of them will cover sorting out individuals. I'll also be sure to read over your input and example when I script it.

Thanks.

Ben Sayer said:
jmurphy said:
So while I understand that for the purposes of your current demos, you didn't want to open this particular can of worms, at some future date, you might want to make a demo showing how Lineascope can help sort out messes like these. It would be a good selling point, IMHO.

That's an excellent idea Jan! I have planned on creating a series of videos about doing evidence-driven genealogy with Ancestry.com. I'll be sure that one of them will cover sorting out individuals. I'll also be sure to read over your input and example when I script it.

Thanks.

My username on Ancestry is packrat74 -- if you'd like to take a look at the specific records I was talking about, send me a note and I'll give you a pointer to the individual in question.

Of course I don't expect any person who isn't a newbie to NEED to collect case studies from other people -- once you've been at this long enough, you find plenty of them on your own. [wink]

I followed this thread earlier in the year with great interest. Now it's coming round to the time of year when I get more time for research.
Like most, by the sounds of it, my recording of sources was not as good as I would like it. So I am thinking of going back and using EM to recheck my family tree, or at least through the main lines.
Before I start, I would like to know if anyone has been using 'Lineascope', 'Bygones' or similar and what their thoughts are?
From the tutorials I prefer 'Lineascope', but like 'jmurphy' I want to know where my data is and have access to it when I want.