ICRC PoW Cards being indexed by FindmyPast

[charlie962]

Have I missed an announcement ? I see from an example here that FindmyPast are starting to copy the ICRC index Cards. Great news if they are planning to do the whole index.

 

Charlie

 

Here is an example of a search result, courtesey FindmyPast.

 

Edited 3 June , 2019 by charlie962

[charlie962]

And I see they are indexing the backing sheets ? eg for Futcher above you get the card and the backing sheet.

Charlie

[David_Blanchard]

Thanks for this.

[charlie962]

It is , of course, dependant on the accuracy of ICRC's records which as anyone who has searched knows, can have some strange spelling etc. However for the first time you can put in a Regiment and get a starting list. Sometimes the date of capture (put in the keyword) can act as a filter. Fantastic potential but...

 

  I have to say I am glad it is FMP doing the transcribing. Normally their attention to accuracy, particularly on Regiments, is excellent.

 

Charlie

[charlie962]

I've just tried a search wth only keyword "21 Mar 1918" and it gives 60531 hits. Obviously this will include index cards, backing sheets and duplicate cards etc and will omit those not correctly indexed but very useful I suspect as a starter list.

 

Note that the month seems to be specified with the three letter abreviation.

 

Charlie

[charlie962]

I did find this on FMP's blog for Fri 24/5/19 . It doesn't explain what they have done so far or are intending to do.

Edited 31 May , 2019 by charlie962

[charlie2]

I did look at one of the records a few weeks ago when there was free access. My impression of what they had transcribed was not at all good, if I recall correctly only one of the available lists had been transcribed or it was just the info on the index card. There were also mistakes contained in the transcription. I did think to myself if thats the best they can do then best forget it. 

 

Charlie2

[charlie2]

1 hour ago, charlie962 said:

And I see they are indexing the backing sheets ? eg for Futcher above you get the card and the backing sheet.

Charlie

 

Both backing sheets? 

[charlie962]

11 minutes ago, charlie2 said:

Both backing sheets? 

Yes, but strangely 2 copies of the Card, one transcribed in more detail than the other, but only one backing sheet ref transcribed each time, if you follow me.

Note the year of Birth. ICRC has 1893 transcribed as 1893. OK. but second sheet has 1898 transcribed as 1896. Oh dear.

Edited 31 May , 2019 by charlie962

[charlie2]

Yes oh dear. How does the other info compare? Cassel list 16.08.1918, transferred from Cassel to Limburg list dated 01.10.1918.

[clk]

Hi,

 

I think that it's a positive development. Accepting there will be mistranscriptions, and errors in the original documents, if you're looking for PoWs from a specific unit rather than just a pre existing known/named individual, the FMP offering seems to add that to their search functionality - it isn't available direct from the ICRC. In those circumstances, maybe the way to use it until we know how it pans out, is to use the FMP search, then given the PA/R references, cross reference to the ICRC records direct.

 

Regards

Chris

Edited 31 May , 2019 by clk

[MaxD]

It may be what Chris means but in order to see the docs on the ICRC site, I take the FMP search then on the ICRC site put any old name (Bloggs Peter works) (sorry Peter if you are watching) which will bring up a random card on which you can use the PA/R numbers from the FMP offering.  Useful cross check of detail and saves wondering how ICRC is going to render Bloggs or the original name.

 

Chris - if that is your system, snap.

 

Max

[clk]

Hi Max,

 

That's the one - . Picked up on a couple of names/details that I hadn't previously had.

 

Regards

Chris

[Skipman]

Just had a look at these. I thought FMP were usually quite good at this but a couple of howlers even from my little town. Fergus Robertson "Fergos Robertson" and John Stewart as "John Etewart"

 

No doubt will be useful though.

 

Mike

[David_Underdown]

As the records are often typed it maybe that they're using OCR - the mistakes highlighted so far would certainly be plausible as OCR errors.

[Skipman]

1 hour ago, David_Underdown said:

As the records are often typed it maybe that they're using OCR - the mistakes highlighted so far would certainly be plausible as OCR errors.

 

So FMP are willing to lose their reputation as being more professional than Ancestry.

 

Mike

[charlie962]

1 hour ago, David_Underdown said:

it maybe that they're using OCR

That might explain why it is not up to what I thought was the usual FMP standard. But I note that with Daily Casualty Lists FMP gave the impression of wanting to rush the job- any search of a man's name / unit / service number brings up far more hits than it should, because FMP produce a hit if all of these appear on the same page of a CasList rather than being for the same man. The Genealogist has a much better definition of search.

 

I hope FMP don't sacrifice quality for quantity.

 

But I do agree it's a great start.

 

Charlie

[David_Underdown]

They're a commercial entity, there's always a level of cost/benefit analysis, both in how you approach transcription (ie manual or automated), how much system development you have to do to try and isolate individual records where many people are listed on the sae page (and I know this one can be a real pain from projects I've been directly involved with), do you try and get something "perfect" to begin with (and to be honest transcription of any sort is never perfect, and the underlying records themselves won't be either), or do you get something out and see how people actually use it, and work out how to improve it as things go along.

[Skipman]

1 hour ago, David_Underdown said:

do you try and get something "perfect" to begin with (and to be honest transcription of any sort is never perfect, and the underlying records themselves won't be either), or do you get something out and see how people actually use it, and work out how to improve it as things go along.

 

Fair enough. There has certainly been a perception tht FMP care(d) more about their indexing. will they go back over it and try correct it?

 

I actually don't have too much trouble finding individuals in the ICRC records, now am used to it. The FMP version could eventually be very good for searching by Reiment etc.

 

Mike

[charlie2]

1 hour ago, Skipman said:

 

Fair enough. There has certainly been a perception tht FMP care(d) more about their indexing. will they go back over it and try correct it?

 

I actually don't have too much trouble finding individuals in the ICRC records, now am used to it. The FMP version could eventually be very good for searching by Reiment etc.

 

Mike

 

I think you are right with your comment that it will be good for searching by regiment but it will only as good as what is in their database.

Does repeating the original mistakes and then making things worse with OCR mistransciptions make for a usable database? And there lies my problem with it all. As anyone who has used the ICRC database knows, mistakes with regiment and name are ten a penny. As you have commented the ICRC database isn‘t that hard to use and the more one uses it the easier it becomes.

 

I would be really interested to know what they make of this example https://grandeguerre.icrc.org/en/File/Details/4026184/3/2/

Charlie2

[David_Underdown]

16 hours ago, Skipman said:

will they go back over it and try correct it?

I'm pretty sure they've made improvements to for example the electoral registers over time.  They're now also doing more with the newspaper collection and highlighting individual stories from that (see their "People in the News" collection).

 

Digitisation plus indexing/transcription is immensely useful but whoever is doing it the end user has to think a bit about the choices that have been made along way and what that means for how you approach searches, and what you might be missing.  There's a lot of academic discussion of this and what it means for research.

 

I'm in the position that I see all sides of the process, with a variety of different companies involved (but as I'm sure you'll understand I can't really pass comment on individual companies).  I also know what a range of surprises a collection can produce during digitisation even when everyone thinks they understand it well, and how sometimes that means you can end up having to make pragmatic decisions how to deal with an issue which you probably would have been able to deal with it differently (ie better) if you'd known about it earlier.

 

Put it this way, an OCRed, directly searchable database will always seem more approachable for casual/inexperienced end users, however, you'll always be able to get more out of a collection if you take the time to understand how the records were originally put together, by whom, and for what purpose (and if you understand some of the limitations of manual and automated transcription).

[charlie962]

18 hours ago, Skipman said:

could eventually be very good for searching by Reiment etc.

Transcription error,  OCR error,  or just one of those finger problems? I wonder how many I'll make in this post. They happen.

 

It is always interesting to see the results of digitisiation of the same record done by different commercial Genealogy search companies:

 

I noted above the difference on Casualty Lists between FindmyPast and TheGenealogist.

 

We all know the variations on Ancestry MIC transcriptions (a whole thread devoted to the more bizarre) compared to FMP, FWR and TheGenealogist. Particularly with unit titles.

 

FWR had a very good transcription of Hospital Admissions but were overtaken by FMP providing access to the backing ledgers together with their wider search choices. As the original registers were handwritten there was no choice but to transcribe rather than use OCR. I'm not aware of any reduction in FMP's 'quality control', (particularly on unit title ?) in that excercise.

 

18 hours ago, Skipman said:

will they go back over it and try correct it?

If that is the intention then FMP would be well advised to say so and avoid unneccesary negative comment!- perhaps they have but I don't know where.

 

I would say that although FMP lack certain records that are on Ancestry, particularly medal rolls and original MICs (both sides), I have always preferred FMP because of their data quality and search capability. I would hate to see this USP spoilt.

 

Charlie

 

 

 

 

 

 

[Keith_history_buff]

Although I wasn't using FMP, but the original ICRC website, I see there are several cards for men with the surname LITTLE yet they are recorded as LYTTLE. I've had a nightmare trying to locate the index card for 8175 Joseph Lythall (PA5888, R51145, R51147).