Second input - shown on Coverage chart


Bernard Sullivan <bms811940@...>
 

My congratulations to all the contributors to freebmd for their work in nearly completing input to 1983. Marriages for December quarter 1969 and deaths September quarter 1980 yet to be completed.

Statistics show that second input of records is growing. About 84 million records. 

As a contributor to the second input I would like to see a colour change on the coverage  charts to show progress for every year that the second input has been completed and there is work on progress. 

Is this a reasonable request?

What do you think?

Is there any support for this request.

Bernard Sullivan 

(The late) John Pain Syndicate




Philip Clarke
 

Thanks for your kind words, Bernard.

 

I’ve done first and second keying (including first-keying post-1983 records) and I am still keying so I’ll willingly accept a small part of your congratulations. You must take some as well for your contributions.

 

Looking at the latest coverage charts, It’s a bit better than that: births and marriages are complete to the end of 1985 and (as you say) deaths up to the end of 1983. With of course the exceptions (or “holes”) that you list below. It also looks likely that the 1985 deaths (not one of mine) could be complete in another month or so, with 1984 deaths, completed perhaps round about the end the year, or early next year.

 

Sadly, at the rate of progress shown by the lighter colours on the coverage charts, it’s also possible that those “holes” might not be filled until after the newer records in the range 1986 to 1992 are completed.

 

Yes, I support your suggestion to those changes to the coverage charts. It would need a change to the programme and then the charts would get updated every month. However, whether that happens and when boils down to the priority given to making it happen.  Getting FreeBMD2 up and running might be regarded as more important and it might get delayed until FreeBMD2 is running with the full set of BMD records.

 

My slight concern is that  from the allocations pages, there appears to be very little in the way of second input after the end of the nineteenth century. To be a bit more precise, after 1900 there is second input for births, marriages and deaths (two out of three cases, but not all three) up to 1904. I can’t see any after that. So, there is about 60 years of double-keying and, if we go up to just 1992, another 90 years of only single-keying.

 

 

Philip Clarke

 

Sent from Mail for Windows 10

 

From: Bernard Sullivan
Sent: 02 August 2020 15:39
To: Advice@freebmd.groups.io
Subject: [FreeBMD-Advice] Second input - shown on Coverage chart

 

My congratulations to all the contributors to freebmd for their work in nearly completing input to 1983. Marriages for December quarter 1969 and deaths September quarter 1980 yet to be completed.

 

Statistics show that second input of records is growing. About 84 million records. 

 

As a contributor to the second input I would like to see a colour change on the coverage  charts to show progress for every year that the second input has been completed and there is work on progress. 

 

Is this a reasonable request?

 

What do you think?

 

Is there any support for this request.

 

Bernard Sullivan 

 

(The late) John Pain Syndicate

 

 

 

 


Dave Marshall
 

A bigger concern than how many records post 1900 haven't yet been double-transcribed is the number of errors in those records (mostly pre-1900 as noted) which have been. I'm currently going through those records where I have been one of the transcribers and I've found an error rate of about 3.5%. So far I've checked about 940,000 records so that's 33,000 errors. I've corrected all of the ones where the errors were mine but can't do anything about the others. This would suggest there could be almost 10 million errors in the database as a whole.

Dave Marshall

On Mon, 3 Aug 2020 at 15:23, Philip Clarke via groups.io <pr.clarke=talktalk.net@groups.io> wrote:

Thanks for your kind words, Bernard.

 

I’ve done first and second keying (including first-keying post-1983 records) and I am still keying so I’ll willingly accept a small part of your congratulations. You must take some as well for your contributions.

 

Looking at the latest coverage charts, It’s a bit better than that: births and marriages are complete to the end of 1985 and (as you say) deaths up to the end of 1983. With of course the exceptions (or “holes”) that you list below. It also looks likely that the 1985 deaths (not one of mine) could be complete in another month or so, with 1984 deaths, completed perhaps round about the end the year, or early next year.

 

Sadly, at the rate of progress shown by the lighter colours on the coverage charts, it’s also possible that those “holes” might not be filled until after the newer records in the range 1986 to 1992 are completed.

 

Yes, I support your suggestion to those changes to the coverage charts. It would need a change to the programme and then the charts would get updated every month. However, whether that happens and when boils down to the priority given to making it happen.  Getting FreeBMD2 up and running might be regarded as more important and it might get delayed until FreeBMD2 is running with the full set of BMD records.

 

My slight concern is that  from the allocations pages, there appears to be very little in the way of second input after the end of the nineteenth century. To be a bit more precise, after 1900 there is second input for births, marriages and deaths (two out of three cases, but not all three) up to 1904. I can’t see any after that. So, there is about 60 years of double-keying and, if we go up to just 1992, another 90 years of only single-keying.

 

 

Philip Clarke

 

Sent from Mail for Windows 10

 

From: Bernard Sullivan
Sent: 02 August 2020 15:39
To: Advice@freebmd.groups.io
Subject: [FreeBMD-Advice] Second input - shown on Coverage chart

 

My congratulations to all the contributors to freebmd for their work in nearly completing input to 1983. Marriages for December quarter 1969 and deaths September quarter 1980 yet to be completed.

 

Statistics show that second input of records is growing. About 84 million records. 

 

As a contributor to the second input I would like to see a colour change on the coverage  charts to show progress for every year that the second input has been completed and there is work on progress. 

 

Is this a reasonable request?

 

What do you think?

 

Is there any support for this request.

 

Bernard Sullivan 

 

(The late) John Pain Syndicate

 

 

 

 


CharlotteBB
 

Hi Dave

There is a method for flagging records that have errors - see the link on the Entry Information screen (marked here in blue on my screen).  When you click on that link a screen with all the fields listed down the page appears, and you can mark the details of the correction for any field that has an error.   You need to note the scan reference for the correct details and put your email and send it through.  I’ve been doing a number of these myself and unfortunately many of them haven’t yet been actioned, but now I add a poster to record what details are wrong and a scan so that the information is recorded.

I receive correction requests occasionally myself and always do them immediately, but possibly original transcribers are no longer transcribing, or their contact details have changed.  

Charlotte Burton-Bell
Wellington NZ.

On 5/08/2020, at 5:53 AM, Dave Marshall <43carnaby@...> wrote:

A bigger concern than how many records post 1900 haven't yet been double-transcribed is the number of errors in those records (mostly pre-1900 as noted) which have been. I'm currently going through those records where I have been one of the transcribers and I've found an error rate of about 3.5%. So far I've checked about 940,000 records so that's 33,000 errors. I've corrected all of the ones where the errors were mine but can't do anything about the others. This would suggest there could be almost 10 million errors in the database as a whole.

Dave Marshall

On Mon, 3 Aug 2020 at 15:23, Philip Clarke via groups.io <pr.clarke=talktalk.net@groups.io> wrote:

Thanks for your kind words, Bernard. 

 

I’ve done first and second keying (including first-keying post-1983 records) and I am still keying so I’ll willingly accept a small part of your congratulations. You must take some as well for your contributions.

 

Looking at the latest coverage charts, It’s a bit better than that: births and marriages are complete to the end of 1985 and (as you say) deaths up to the end of 1983. With of course the exceptions (or “holes”) that you list below. It also looks likely that the 1985 deaths (not one of mine) could be complete in another month or so, with 1984 deaths, completed perhaps round about the end the year, or early next year.

 

Sadly, at the rate of progress shown by the lighter colours on the coverage charts, it’s also possible that those “holes” might not be filled until after the newer records in the range 1986 to 1992 are completed.

 

Yes, I support your suggestion to those changes to the coverage charts. It would need a change to the programme and then the charts would get updated every month. However, whether that happens and when boils down to the priority given to making it happen.  Getting FreeBMD2 up and running might be regarded as more important and it might get delayed until FreeBMD2 is running with the full set of BMD records.

 

My slight concern is that  from the allocations pages, there appears to be very little in the way of second input after the end of the nineteenth century. To be a bit more precise, after 1900 there is second input for births, marriages and deaths (two out of three cases, but not all three) up to 1904. I can’t see any after that. So, there is about 60 years of double-keying and, if we go up to just 1992, another 90 years of only single-keying. 

 

 

Philip Clarke

 

Sent from Mail for Windows 10

 

From: Bernard Sullivan
Sent: 02 August 2020 15:39
To: Advice@freebmd.groups.io
Subject: [FreeBMD-Advice] Second input - shown on Coverage chart

 

My congratulations to all the contributors to freebmd for their work in nearly completing input to 1983. Marriages for December quarter 1969 and deaths September quarter 1980 yet to be completed.

 

Statistics show that second input of records is growing. About 84 million records. 

 

As a contributor to the second input I would like to see a colour change on the coverage  charts to show progress for every year that the second input has been completed and there is work on progress. 

 

Is this a reasonable request?

 

What do you think?

 

Is there any support for this request.

 

Bernard Sullivan 

 

(The late) John Pain Syndicate

 

 

 

 





Jean Flannery
 

Hi Charlotte

Thanks for that and good to see someone else from NZ! 

Jean Flannery
Christchurch NZ

On 5/08/2020, at 11:03 AM, Family Bell/Burton-Bell <ccnbells@...> wrote:

Hi Dave

There is a method for flagging records that have errors - see the link on the Entry Information screen (marked here in blue on my screen).  When you click on that link a screen with all the fields listed down the page appears, and you can mark the details of the correction for any field that has an error.   You need to note the scan reference for the correct details and put your email and send it through.  I’ve been doing a number of these myself and unfortunately many of them haven’t yet been actioned, but now I add a poster to record what details are wrong and a scan so that the information is recorded.

I receive correction requests occasionally myself and always do them immediately, but possibly original transcribers are no longer transcribing, or their contact details have changed.  

Charlotte Burton-Bell
Wellington NZ.

On 5/08/2020, at 5:53 AM, Dave Marshall <43carnaby@...> wrote:

A bigger concern than how many records post 1900 haven't yet been double-transcribed is the number of errors in those records (mostly pre-1900 as noted) which have been. I'm currently going through those records where I have been one of the transcribers and I've found an error rate of about 3.5%. So far I've checked about 940,000 records so that's 33,000 errors. I've corrected all of the ones where the errors were mine but can't do anything about the others. This would suggest there could be almost 10 million errors in the database as a whole.

Dave Marshall

On Mon, 3 Aug 2020 at 15:23, Philip Clarke via groups.io <pr.clarke=talktalk.net@groups.io> wrote:

Thanks for your kind words, Bernard. 

 

I’ve done first and second keying (including first-keying post-1983 records) and I am still keying so I’ll willingly accept a small part of your congratulations. You must take some as well for your contributions.

 

Looking at the latest coverage charts, It’s a bit better than that: births and marriages are complete to the end of 1985 and (as you say) deaths up to the end of 1983. With of course the exceptions (or “holes”) that you list below. It also looks likely that the 1985 deaths (not one of mine) could be complete in another month or so, with 1984 deaths, completed perhaps round about the end the year, or early next year.

 

Sadly, at the rate of progress shown by the lighter colours on the coverage charts, it’s also possible that those “holes” might not be filled until after the newer records in the range 1986 to 1992 are completed.

 

Yes, I support your suggestion to those changes to the coverage charts. It would need a change to the programme and then the charts would get updated every month. However, whether that happens and when boils down to the priority given to making it happen.  Getting FreeBMD2 up and running might be regarded as more important and it might get delayed until FreeBMD2 is running with the full set of BMD records.

 

My slight concern is that  from the allocations pages, there appears to be very little in the way of second input after the end of the nineteenth century. To be a bit more precise, after 1900 there is second input for births, marriages and deaths (two out of three cases, but not all three) up to 1904. I can’t see any after that. So, there is about 60 years of double-keying and, if we go up to just 1992, another 90 years of only single-keying. 

 

 

Philip Clarke

 

Sent from Mail for Windows 10

 

From: Bernard Sullivan
Sent: 02 August 2020 15:39
To: Advice@freebmd.groups.io
Subject: [FreeBMD-Advice] Second input - shown on Coverage chart

 

My congratulations to all the contributors to freebmd for their work in nearly completing input to 1983. Marriages for December quarter 1969 and deaths September quarter 1980 yet to be completed.

 

Statistics show that second input of records is growing. About 84 million records. 

 

As a contributor to the second input I would like to see a colour change on the coverage  charts to show progress for every year that the second input has been completed and there is work on progress. 

 

Is this a reasonable request?

 

What do you think?

 

Is there any support for this request.

 

Bernard Sullivan 

 

(The late) John Pain Syndicate

 

 

 

 





Philip Clarke
 

There was some quite heated discussions about this topic a few years ago in respect of comparing “A” and “B” records.

 

Some of the early images were of poor quality so in some cases a lot of UCF’s being used, sometimes 10% or more. I’m not sure how that is measured: 10 % of the individual records on a page with one or more UCF(s) or 10% of the characters on a page were UCF’s? Those transcriptions were being “retired” as better images were being put onto the system, but the original transcriber kept the “recognition of the work done”. There are other errors, of the type we are discussing here: errors in transcriptions and missing entries.

 

The system laid down by FreeBMD was the each record would be typed independently by two different transcribers and both would be put onto the system.

 

Several years ago an un-named transcriber was transcribing second-entries and then having done so was comparing the “A” and the “B” records against each other. Note: the early pages have 375 records per page and they are double pages; the pages shrank over time, to about 150 lines (max) per page and are now 100 lines (max) per page.

 

If the “A” record was wrong the second-entry-transcriber would then raise correction requests and at that time the correction system was one person: no longer with us, but not forgotten – his syndicate still remains active, I believe.  Dave’s figures below seem to suggest an error rate of one in thirty: so that is about 25 correction requests for each double 375 record page. That transcriber was asked to stop doing this as it appeared to compromise the independent-transcriptions process. The alternative argument was that the “B” transcriber was a public-mined person who was improving the quality of the records. Independent transcribing with independent checking at a later date “won” the argument.

 

Perhaps, there is a checking system in place and that is what you are doing? The discussions concluded all those years ago with an aim that a post of Checker in chief (those are my words, not the words of FreeBMD) would be appointed to manage this. A lot of people offered their services. There does not appear to be any publicity  (perhaps I missed it) that this system is up and running, or is due to happen sooner or later.

 

Sorry, if this sound somewhat depressing.

 

Philip

 

Sent from Mail for Windows 10

 

From: Family Bell/Burton-Bell
Sent: 05 August 2020 00:03
To: Advice@freebmd.groups.io; Dave Marshall
Subject: Re: [FreeBMD-Advice] Second input - shown on Coverage chart

 

Hi Dave

 

There is a method for flagging records that have errors - see the link on the Entry Information screen (marked here in blue on my screen).  When you click on that link a screen with all the fields listed down the page appears, and you can mark the details of the correction for any field that has an error.   You need to note the scan reference for the correct details and put your email and send it through.  I’ve been doing a number of these myself and unfortunately many of them haven’t yet been actioned, but now I add a poster to record what details are wrong and a scan so that the information is recorded.

 

I receive correction requests occasionally myself and always do them immediately, but possibly original transcribers are no longer transcribing, or their contact details have changed.  

 

Charlotte Burton-Bell

Wellington NZ.



On 5/08/2020, at 5:53 AM, Dave Marshall <43carnaby@...> wrote:

 

A bigger concern than how many records post 1900 haven't yet been double-transcribed is the number of errors in those records (mostly pre-1900 as noted) which have been. I'm currently going through those records where I have been one of the transcribers and I've found an error rate of about 3.5%. So far I've checked about 940,000 records so that's 33,000 errors. I've corrected all of the ones where the errors were mine but can't do anything about the others. This would suggest there could be almost 10 million errors in the database as a whole.

 

Dave Marshall

 

On Mon, 3 Aug 2020 at 15:23, Philip Clarke via groups.io <pr.clarke=talktalk.net@groups.io> wrote:

Thanks for your kind words, Bernard. 

 

I’ve done first and second keying (including first-keying post-1983 records) and I am still keying so I’ll willingly accept a small part of your congratulations. You must take some as well for your contributions.

 

Looking at the latest coverage charts, It’s a bit better than that: births and marriages are complete to the end of 1985 and (as you say) deaths up to the end of 1983. With of course the exceptions (or “holes”) that you list below. It also looks likely that the 1985 deaths (not one of mine) could be complete in another month or so, with 1984 deaths, completed perhaps round about the end the year, or early next year.

 

Sadly, at the rate of progress shown by the lighter colours on the coverage charts, it’s also possible that those “holes” might not be filled until after the newer records in the range 1986 to 1992 are completed.

 

Yes, I support your suggestion to those changes to the coverage charts. It would need a change to the programme and then the charts would get updated every month. However, whether that happens and when boils down to the priority given to making it happen.  Getting FreeBMD2 up and running might be regarded as more important and it might get delayed until FreeBMD2 is running with the full set of BMD records.

 

My slight concern is that  from the allocations pages, there appears to be very little in the way of second input after the end of the nineteenth century. To be a bit more precise, after 1900 there is second input for births, marriages and deaths (two out of three cases, but not all three) up to 1904. I can’t see any after that. So, there is about 60 years of double-keying and, if we go up to just 1992, another 90 years of only single-keying. 

 

 

Philip Clarke

 

Sent from Mail for Windows 10

 

From: Bernard Sullivan
Sent: 02 August 2020 15:39
To: Advice@freebmd.groups.io
Subject: [FreeBMD-Advice] Second input - shown on Coverage chart

 

My congratulations to all the contributors to freebmd for their work in nearly completing input to 1983. Marriages for December quarter 1969 and deaths September quarter 1980 yet to be completed.

 

Statistics show that second input of records is growing. About 84 million records. 

 

As a contributor to the second input I would like to see a colour change on the coverage  charts to show progress for every year that the second input has been completed and there is work on progress. 

 

Is this a reasonable request?

 

What do you think?

 

Is there any support for this request.

 

Bernard Sullivan 

 

(The late) John Pain Syndicate

 

 

 

 

 

 

 

 


'Barrie'
 

The purpose of the coverage charts is to enable researchers to determine if the periods they are interested in have been fully transcribed. The coverage charts are linked to from the home page to remind researchers of this. There is hence a clear and definite need for this information.

In contrast adding the second keying is a "nice to know" for far fewer people, so is not a work package that is likely to get resources.

The second keying can be estimated by using the Entry Counts data with the proportion double keyed given by

(Total - Unique) / Unique

although it will be found that this gives figures that indicate it is only a very rough estimate.

Regards

Barrie

On 02/08/2020 14:30, Bernard Sullivan wrote:
My congratulations to all the contributors to freebmd for their work in nearly completing input to 1983. Marriages for December quarter 1969 and deaths September quarter 1980 yet to be completed.

Statistics show that second input of records is growing. About 84 million records. 

As a contributor to the second input I would like to see a colour change on the coverage  charts to show progress for every year that the second input has been completed and there is work on progress. 

Is this a reasonable request?

What do you think?

Is there any support for this request.

Bernard Sullivan 

(The late) John Pain Syndicate





--Certified Virus Free by 4SecureMail.com ICSA-Certified Scanner--


'Barrie'
 

I don't recall the "heated discussion" but the policy is made quite clear in the FreeBMD process and it states that resolution of different transcriptions may only be done by someone not connected with either of the syndicates doing the transcribing. Equally the conditions for submitting a correction make it clear that corrections must not be used to submit "resolution corrections".

Regards

Barrie

On 05/08/2020 07:29, Philip Clarke via groups.io wrote:

There was some quite heated discussions about this topic a few years ago in respect of comparing “A” and “B” records.

 

Some of the early images were of poor quality so in some cases a lot of UCF’s being used, sometimes 10% or more. I’m not sure how that is measured: 10 % of the individual records on a page with one or more UCF(s) or 10% of the characters on a page were UCF’s? Those transcriptions were being “retired” as better images were being put onto the system, but the original transcriber kept the “recognition of the work done”. There are other errors, of the type we are discussing here: errors in transcriptions and missing entries.

 

The system laid down by FreeBMD was the each record would be typed independently by two different transcribers and both would be put onto the system.

 

Several years ago an un-named transcriber was transcribing second-entries and then having done so was comparing the “A” and the “B” records against each other. Note: the early pages have 375 records per page and they are double pages; the pages shrank over time, to about 150 lines (max) per page and are now 100 lines (max) per page.

 

If the “A” record was wrong the second-entry-transcriber would then raise correction requests and at that time the correction system was one person: no longer with us, but not forgotten – his syndicate still remains active, I believe.  Dave’s figures below seem to suggest an error rate of one in thirty: so that is about 25 correction requests for each double 375 record page. That transcriber was asked to stop doing this as it appeared to compromise the independent-transcriptions process. The alternative argument was that the “B” transcriber was a public-mined person who was improving the quality of the records. Independent transcribing with independent checking at a later date “won” the argument.

 

Perhaps, there is a checking system in place and that is what you are doing? The discussions concluded all those years ago with an aim that a post of Checker in chief (those are my words, not the words of FreeBMD) would be appointed to manage this. A lot of people offered their services. There does not appear to be any publicity  (perhaps I missed it) that this system is up and running, or is due to happen sooner or later.

 

Sorry, if this sound somewhat depressing.

 

Philip

 

Sent from Mail for Windows 10

 

From: Family Bell/Burton-Bell
Sent: 05 August 2020 00:03
To: Advice@freebmd.groups.io; Dave Marshall
Subject: Re: [FreeBMD-Advice] Second input - shown on Coverage chart

 

Hi Dave

 

There is a method for flagging records that have errors - see the link on the Entry Information screen (marked here in blue on my screen).  When you click on that link a screen with all the fields listed down the page appears, and you can mark the details of the correction for any field that has an error.   You need to note the scan reference for the correct details and put your email and send it through.  I’ve been doing a number of these myself and unfortunately many of them haven’t yet been actioned, but now I add a poster to record what details are wrong and a scan so that the information is recorded.

 

I receive correction requests occasionally myself and always do them immediately, but possibly original transcribers are no longer transcribing, or their contact details have changed.  

 

Charlotte Burton-Bell

Wellington NZ.



On 5/08/2020, at 5:53 AM, Dave Marshall <43carnaby@...> wrote:

 

A bigger concern than how many records post 1900 haven't yet been double-transcribed is the number of errors in those records (mostly pre-1900 as noted) which have been. I'm currently going through those records where I have been one of the transcribers and I've found an error rate of about 3.5%. So far I've checked about 940,000 records so that's 33,000 errors. I've corrected all of the ones where the errors were mine but can't do anything about the others. This would suggest there could be almost 10 million errors in the database as a whole.

 

Dave Marshall

 

On Mon, 3 Aug 2020 at 15:23, Philip Clarke via groups.io <pr.clarke=talktalk.net@groups.io> wrote:

Thanks for your kind words, Bernard. 

 

I’ve done first and second keying (including first-keying post-1983 records) and I am still keying so I’ll willingly accept a small part of your congratulations. You must take some as well for your contributions.

 

Looking at the latest coverage charts, It’s a bit better than that: births and marriages are complete to the end of 1985 and (as you say) deaths up to the end of 1983. With of course the exceptions (or “holes”) that you list below. It also looks likely that the 1985 deaths (not one of mine) could be complete in another month or so, with 1984 deaths, completed perhaps round about the end the year, or early next year.

 

Sadly, at the rate of progress shown by the lighter colours on the coverage charts, it’s also possible that those “holes” might not be filled until after the newer records in the range 1986 to 1992 are completed.

 

Yes, I support your suggestion to those changes to the coverage charts. It would need a change to the programme and then the charts would get updated every month. However, whether that happens and when boils down to the priority given to making it happen.  Getting FreeBMD2 up and running might be regarded as more important and it might get delayed until FreeBMD2 is running with the full set of BMD records.

 

My slight concern is that  from the allocations pages, there appears to be very little in the way of second input after the end of the nineteenth century. To be a bit more precise, after 1900 there is second input for births, marriages and deaths (two out of three cases, but not all three) up to 1904. I can’t see any after that. So, there is about 60 years of double-keying and, if we go up to just 1992, another 90 years of only single-keying. 

 

 

Philip Clarke

 

Sent from Mail for Windows 10

 

From: Bernard Sullivan
Sent: 02 August 2020 15:39
To: Advice@freebmd.groups.io
Subject: [FreeBMD-Advice] Second input - shown on Coverage chart

 

My congratulations to all the contributors to freebmd for their work in nearly completing input to 1983. Marriages for December quarter 1969 and deaths September quarter 1980 yet to be completed.

 

Statistics show that second input of records is growing. About 84 million records. 

 

As a contributor to the second input I would like to see a colour change on the coverage  charts to show progress for every year that the second input has been completed and there is work on progress. 

 

Is this a reasonable request?

 

What do you think?

 

Is there any support for this request.

 

Bernard Sullivan 

 

(The late) John Pain Syndicate

 

 

 

 

 

 

 

 



--Certified Virus Free by 4SecureMail.com ICSA-Certified Scanner--


Dave Marshall
 

Barrie,

What does the policy say about checking one's own transcriptions for errors not picked up at the time of submission?

Dave

On Thu, 6 Aug 2020 at 21:14, Barrie <freeukgen@...> wrote:

I don't recall the "heated discussion" but the policy is made quite clear in the FreeBMD process and it states that resolution of different transcriptions may only be done by someone not connected with either of the syndicates doing the transcribing. Equally the conditions for submitting a correction make it clear that corrections must not be used to submit "resolution corrections".

Regards

Barrie

On 05/08/2020 07:29, Philip Clarke via groups.io wrote:

There was some quite heated discussions about this topic a few years ago in respect of comparing “A” and “B” records.

 

Some of the early images were of poor quality so in some cases a lot of UCF’s being used, sometimes 10% or more. I’m not sure how that is measured: 10 % of the individual records on a page with one or more UCF(s) or 10% of the characters on a page were UCF’s? Those transcriptions were being “retired” as better images were being put onto the system, but the original transcriber kept the “recognition of the work done”. There are other errors, of the type we are discussing here: errors in transcriptions and missing entries.

 

The system laid down by FreeBMD was the each record would be typed independently by two different transcribers and both would be put onto the system.

 

Several years ago an un-named transcriber was transcribing second-entries and then having done so was comparing the “A” and the “B” records against each other. Note: the early pages have 375 records per page and they are double pages; the pages shrank over time, to about 150 lines (max) per page and are now 100 lines (max) per page.

 

If the “A” record was wrong the second-entry-transcriber would then raise correction requests and at that time the correction system was one person: no longer with us, but not forgotten – his syndicate still remains active, I believe.  Dave’s figures below seem to suggest an error rate of one in thirty: so that is about 25 correction requests for each double 375 record page. That transcriber was asked to stop doing this as it appeared to compromise the independent-transcriptions process. The alternative argument was that the “B” transcriber was a public-mined person who was improving the quality of the records. Independent transcribing with independent checking at a later date “won” the argument.

 

Perhaps, there is a checking system in place and that is what you are doing? The discussions concluded all those years ago with an aim that a post of Checker in chief (those are my words, not the words of FreeBMD) would be appointed to manage this. A lot of people offered their services. There does not appear to be any publicity  (perhaps I missed it) that this system is up and running, or is due to happen sooner or later.

 

Sorry, if this sound somewhat depressing.

 

Philip

 

Sent from Mail for Windows 10

 

From: Family Bell/Burton-Bell
Sent: 05 August 2020 00:03
To: Advice@freebmd.groups.io; Dave Marshall
Subject: Re: [FreeBMD-Advice] Second input - shown on Coverage chart

 

Hi Dave

 

There is a method for flagging records that have errors - see the link on the Entry Information screen (marked here in blue on my screen).  When you click on that link a screen with all the fields listed down the page appears, and you can mark the details of the correction for any field that has an error.   You need to note the scan reference for the correct details and put your email and send it through.  I’ve been doing a number of these myself and unfortunately many of them haven’t yet been actioned, but now I add a poster to record what details are wrong and a scan so that the information is recorded.

 

I receive correction requests occasionally myself and always do them immediately, but possibly original transcribers are no longer transcribing, or their contact details have changed.  

 

Charlotte Burton-Bell

Wellington NZ.



On 5/08/2020, at 5:53 AM, Dave Marshall <43carnaby@...> wrote:

 

A bigger concern than how many records post 1900 haven't yet been double-transcribed is the number of errors in those records (mostly pre-1900 as noted) which have been. I'm currently going through those records where I have been one of the transcribers and I've found an error rate of about 3.5%. So far I've checked about 940,000 records so that's 33,000 errors. I've corrected all of the ones where the errors were mine but can't do anything about the others. This would suggest there could be almost 10 million errors in the database as a whole.

 

Dave Marshall

 

On Mon, 3 Aug 2020 at 15:23, Philip Clarke via groups.io <pr.clarke=talktalk.net@groups.io> wrote:

Thanks for your kind words, Bernard. 

 

I’ve done first and second keying (including first-keying post-1983 records) and I am still keying so I’ll willingly accept a small part of your congratulations. You must take some as well for your contributions.

 

Looking at the latest coverage charts, It’s a bit better than that: births and marriages are complete to the end of 1985 and (as you say) deaths up to the end of 1983. With of course the exceptions (or “holes”) that you list below. It also looks likely that the 1985 deaths (not one of mine) could be complete in another month or so, with 1984 deaths, completed perhaps round about the end the year, or early next year.

 

Sadly, at the rate of progress shown by the lighter colours on the coverage charts, it’s also possible that those “holes” might not be filled until after the newer records in the range 1986 to 1992 are completed.

 

Yes, I support your suggestion to those changes to the coverage charts. It would need a change to the programme and then the charts would get updated every month. However, whether that happens and when boils down to the priority given to making it happen.  Getting FreeBMD2 up and running might be regarded as more important and it might get delayed until FreeBMD2 is running with the full set of BMD records.

 

My slight concern is that  from the allocations pages, there appears to be very little in the way of second input after the end of the nineteenth century. To be a bit more precise, after 1900 there is second input for births, marriages and deaths (two out of three cases, but not all three) up to 1904. I can’t see any after that. So, there is about 60 years of double-keying and, if we go up to just 1992, another 90 years of only single-keying. 

 

 

Philip Clarke

 

Sent from Mail for Windows 10

 

From: Bernard Sullivan
Sent: 02 August 2020 15:39
To: Advice@freebmd.groups.io
Subject: [FreeBMD-Advice] Second input - shown on Coverage chart

 

My congratulations to all the contributors to freebmd for their work in nearly completing input to 1983. Marriages for December quarter 1969 and deaths September quarter 1980 yet to be completed.

 

Statistics show that second input of records is growing. About 84 million records. 

 

As a contributor to the second input I would like to see a colour change on the coverage  charts to show progress for every year that the second input has been completed and there is work on progress. 

 

Is this a reasonable request?

 

What do you think?

 

Is there any support for this request.

 

Bernard Sullivan 

 

(The late) John Pain Syndicate

 

 

 

 

 

 

 

 



--Certified Virus Free by 4SecureMail.com ICSA-Certified Scanner--


Jeff Coleman
 

In response to Dave’s query, whenever I get a request for corrections to one of my files, or the files of former transcribers that I deal with for corrections purposes, I try to go through with the best available image and correct UCF entries where possible. Occasionally when dealing with transcriptions originally done from the very poor UKD-01 scans I check all the page numbers and sometimes everything on the scan. I encourage transcribers who receive correction requests to have a look at other entries in the file, such as UCF items, to see whether they can easily correct them. Some of my team have been going systematically through the pages they did around 2004-2005 looing to improve on the transcriptions from the better scan images now available.

 

However there is no obligation or recommendation as far as I know.

 

Jeff Coleman

 

Jeff.Coleman@...

 

Syndicate Co-ordinator of the Jeff Coleman Syndicate of volunteer transcribers working for FreeBMD

www.FreeBMD.org.uk

 

 

 

 


'Barrie'
 

I don't have any further information over and above what is given in the links I included.

Regards

Barrie

On 06/08/2020 22:26, Dave Marshall wrote:
Barrie,

What does the policy say about checking one's own transcriptions for errors not picked up at the time of submission?

Dave

On Thu, 6 Aug 2020 at 21:14, Barrie <freeukgen@...> wrote:

I don't recall the "heated discussion" but the policy is made quite clear in the FreeBMD process and it states that resolution of different transcriptions may only be done by someone not connected with either of the syndicates doing the transcribing. Equally the conditions for submitting a correction make it clear that corrections must not be used to submit "resolution corrections".

Regards

Barrie

On 05/08/2020 07:29, Philip Clarke via groups.io wrote:

There was some quite heated discussions about this topic a few years ago in respect of comparing “A” and “B” records.

 

Some of the early images were of poor quality so in some cases a lot of UCF’s being used, sometimes 10% or more. I’m not sure how that is measured: 10 % of the individual records on a page with one or more UCF(s) or 10% of the characters on a page were UCF’s? Those transcriptions were being “retired” as better images were being put onto the system, but the original transcriber kept the “recognition of the work done”. There are other errors, of the type we are discussing here: errors in transcriptions and missing entries.

 

The system laid down by FreeBMD was the each record would be typed independently by two different transcribers and both would be put onto the system.

 

Several years ago an un-named transcriber was transcribing second-entries and then having done so was comparing the “A” and the “B” records against each other. Note: the early pages have 375 records per page and they are double pages; the pages shrank over time, to about 150 lines (max) per page and are now 100 lines (max) per page.

 

If the “A” record was wrong the second-entry-transcriber would then raise correction requests and at that time the correction system was one person: no longer with us, but not forgotten – his syndicate still remains active, I believe.  Dave’s figures below seem to suggest an error rate of one in thirty: so that is about 25 correction requests for each double 375 record page. That transcriber was asked to stop doing this as it appeared to compromise the independent-transcriptions process. The alternative argument was that the “B” transcriber was a public-mined person who was improving the quality of the records. Independent transcribing with independent checking at a later date “won” the argument.

 

Perhaps, there is a checking system in place and that is what you are doing? The discussions concluded all those years ago with an aim that a post of Checker in chief (those are my words, not the words of FreeBMD) would be appointed to manage this. A lot of people offered their services. There does not appear to be any publicity  (perhaps I missed it) that this system is up and running, or is due to happen sooner or later.

 

Sorry, if this sound somewhat depressing.

 

Philip

 

Sent from Mail for Windows 10

 

From: Family Bell/Burton-Bell
Sent: 05 August 2020 00:03
To: Advice@freebmd.groups.io; Dave Marshall
Subject: Re: [FreeBMD-Advice] Second input - shown on Coverage chart

 

Hi Dave

 

There is a method for flagging records that have errors - see the link on the Entry Information screen (marked here in blue on my screen).  When you click on that link a screen with all the fields listed down the page appears, and you can mark the details of the correction for any field that has an error.   You need to note the scan reference for the correct details and put your email and send it through.  I’ve been doing a number of these myself and unfortunately many of them haven’t yet been actioned, but now I add a poster to record what details are wrong and a scan so that the information is recorded.

 

I receive correction requests occasionally myself and always do them immediately, but possibly original transcribers are no longer transcribing, or their contact details have changed.  

 

Charlotte Burton-Bell

Wellington NZ.



On 5/08/2020, at 5:53 AM, Dave Marshall <43carnaby@...> wrote:

 

A bigger concern than how many records post 1900 haven't yet been double-transcribed is the number of errors in those records (mostly pre-1900 as noted) which have been. I'm currently going through those records where I have been one of the transcribers and I've found an error rate of about 3.5%. So far I've checked about 940,000 records so that's 33,000 errors. I've corrected all of the ones where the errors were mine but can't do anything about the others. This would suggest there could be almost 10 million errors in the database as a whole.

 

Dave Marshall

 

On Mon, 3 Aug 2020 at 15:23, Philip Clarke via groups.io <pr.clarke=talktalk.net@groups.io> wrote:

Thanks for your kind words, Bernard. 

 

I’ve done first and second keying (including first-keying post-1983 records) and I am still keying so I’ll willingly accept a small part of your congratulations. You must take some as well for your contributions.

 

Looking at the latest coverage charts, It’s a bit better than that: births and marriages are complete to the end of 1985 and (as you say) deaths up to the end of 1983. With of course the exceptions (or “holes”) that you list below. It also looks likely that the 1985 deaths (not one of mine) could be complete in another month or so, with 1984 deaths, completed perhaps round about the end the year, or early next year.

 

Sadly, at the rate of progress shown by the lighter colours on the coverage charts, it’s also possible that those “holes” might not be filled until after the newer records in the range 1986 to 1992 are completed.

 

Yes, I support your suggestion to those changes to the coverage charts. It would need a change to the programme and then the charts would get updated every month. However, whether that happens and when boils down to the priority given to making it happen.  Getting FreeBMD2 up and running might be regarded as more important and it might get delayed until FreeBMD2 is running with the full set of BMD records.

 

My slight concern is that  from the allocations pages, there appears to be very little in the way of second input after the end of the nineteenth century. To be a bit more precise, after 1900 there is second input for births, marriages and deaths (two out of three cases, but not all three) up to 1904. I can’t see any after that. So, there is about 60 years of double-keying and, if we go up to just 1992, another 90 years of only single-keying. 

 

 

Philip Clarke

 

Sent from Mail for Windows 10

 

From: Bernard Sullivan
Sent: 02 August 2020 15:39
To: Advice@freebmd.groups.io
Subject: [FreeBMD-Advice] Second input - shown on Coverage chart

 

My congratulations to all the contributors to freebmd for their work in nearly completing input to 1983. Marriages for December quarter 1969 and deaths September quarter 1980 yet to be completed.

 

Statistics show that second input of records is growing. About 84 million records. 

 

As a contributor to the second input I would like to see a colour change on the coverage  charts to show progress for every year that the second input has been completed and there is work on progress. 

 

Is this a reasonable request?

 

What do you think?

 

Is there any support for this request.

 

Bernard Sullivan 

 

(The late) John Pain Syndicate

 

 

 

 

 

 

 

 



--Certified Virus Free by 4SecureMail.com ICSA-Certified Scanner--




--Certified Virus Free by 4SecureMail.com ICSA-Certified Scanner--