Project Update – teaching a computer to READ Bentham

By Louise Seaward, on 9 June 2017

The difficulty of Bentham’s handwriting is notorious.  At the Bentham Project, we have years of experience of transcribing Bentham but you will still regularly find us hunched over a manuscript with a magnifying glass or blankly staring at a digital image on a computer screen, zooming in and out on a particular word.

One of the Bentham Project's favourite tools

One of the Bentham Project’s favourite tools

 

Across the last few years, we have been working closely with various teams of computer scientists in the hope of making progress on the automated recognition of Bentham’s writing.  This collaboration started under the tranScriptorium project in 2013 and now continues in its successor project READ (Recognition and Enrichment of Archival Documents).

READ’s mission is to make archival collections more accessible through the development and dissemination of Handwritten Text Recognition (HTR) technology.  This technology is freely available through the Transkribus platform.  Using algorithms of machine learning, it is possible to teach a computer to read a particular style of writing.  The technology is trained by being shown images of documents and their accurate transcriptions.  Anyone can start a test project with around 20,000 words or around 100 pages.

Under the tranScriptorium project, we initially had some success in training a model to process manuscripts from the Bentham collection.  Using around 900 pages of Bentham images and transcripts, researchers from the Pattern Recognition and Human Language Technology (PRHLT) research centre at the Universitat Politècnica de València created a HTR model for us using statistical algorithms called Hidden Markov Models.  This model was able to produce relatively accurate transcriptions of the Bentham papers, with a Character Error Rate of around 18% (meaning that around 82% of the characters in a transcript would be correct).

In the first stage of the READ project, we have already been able to enhance the accuracy of the HTR technology.  The team at the Computational Intelligence Technology Lab (CITlab) at the University of Rostock created a new model using this same dataset.  This model was based on Neural Networks, computational models for machine learning which work similarly to the human brain.  This model can produce automatic transcripts of the Bentham papers with a Character Error Rate of only 5-10%.

Now it’s time to take things up a notch!  In our first experiments with HTR, we put forward ‘easier’ documents for the computer to process.  These tended to be pages written by Bentham’s secretaries where the layout is clear and the handwriting relatively neat.  Now we want to test how the computer copes with some of the worst examples of Bentham’s writing.  We are producing a new set of training data based on a selection of manuscripts which were written by Bentham himself when the philosopher was in his eighties.  Box xxx of the Bentham Papers in UCL Special Collections contains the Blackstone Familiarized papers.  These were part of Bentham’s lifelong obsession with critiquing the work of William Blackstone, the English jurist who was most famous for his Commentaries on the Laws of England (1765-9).  Bentham first turned against Blackstone as a teenage student when he attended his lectures at the University of Oxford.  In several published works and unpublished papers, Bentham argued that Blackstone was an apologist for the obvious inadequacies in the English legal system and blind to the necessity of reform.

 

Screenshot of page from Blackstone Familiarized in Transkribus. UCL Special Collections, Bentham Papers, Box xxx, fo. 156 [Image: UCL Special Collections]

Screenshot of page from Blackstone Familiarized in Transkribus. UCL Special Collections, Bentham Papers, Box xxx, fo. 156 [Image: UCL Special Collections]

The Blackstone Familiarized papers have been digitised by UCL Creative Media Services and transcribed by Professor Philip Schofield, the Director of the Bentham Project and General Editor of the Collected Works of Jeremy Bentham.  The images were uploaded to Transkribus and Chris Riley, a PhD student from the Faculty of Laws, has been marking the lines of text on each image and then copying the transcripts into the platform.

We are aiming to produce 200 pages of ‘difficult’ Bentham training data which can be fed into a new version of our latest HTR model.  We are also interested in comparing the accuracy of different models.  How far does this new material enhance the accuracy of the models we already have and would it be worthwhile to have separate models for Bentham himself and his secretaries?

The prospect of the automated recognition of Bentham’s handwriting would considerably speed up the full transcription of Bentham’s writings and the publication of his Collected Works.  We also want to experiment with HTR technology in a new version of Transcribe Bentham where volunteer transcribers could ask the computer to provide suggested readings of words that they are difficult to decipher.  Until then, we have some more transcribing to do!

Changing up our blog schedule!

By Louise Seaward, on 28 May 2017

Hello readers!  Another month has flown by with a flurry of activity from our volunteer transcribers.  As the month comes to a close, we are rolling out some changes to our blog schedule.  There will now be at least two posts each month on the Transcribe Bentham blog.

  1. Transcription Update.  Our familar statistics, detailing what was transcribed and when, will become a monthly feature.  The Transcribe Bentham team will still be keeping an eye on things every week for our own records, so you can get in touch at any time if you want more information on our latest statistics.
  2. Project Update.  We will be adding at least one new post every month to shed light on the activities of our volunteers and the research and editorial work undertaken by the Bentham Project.  Expect to hear more about the intricacies of crowdsourcing, new discoveries from Bentham’s papers and our experiments with training a computer to recognise Bentham’s handwriting as part of the READ project!

We hope this new format will be interesting and help us in our mission to spread the word about Bentham far and wide!  If you have any ideas for topics that you would like to see discussed on the blog, please feel free to leave a comment or send us an email.

Transcription Update – 5 to 26 May 2017

By Louise Seaward, on 28 May 2017

Hello everyone!  We’re here with the latest transcription update to share how many pages have now been transcribed.  As usual, we need to start by thanking everyone who has been transcribing recently.

We need to let you know that these statistics updates will now appear monthly, rather than weekly on our blog – check on our latest post for more information on this change!

These are the latest statistics as of 26 May 2017.

18,355 manuscript pages have now been transcribed or partially-transcribed.  Of these transcripts, 17,301 (94%) have been checked and approved by TB staff.

Over the past four weeks, volunteers have worked on a total of 173 manuscript pages.  This means that an average of 43 pages have been transcribed each week during the past month.

The more detailed progress chart is as follows:

Box No. of manuscripts worked on No. of manuscripts in box Completion
Box 1 769 794 96%
Box 2 729 753 96%
Box 4 45 694 6%
Box 5 201 290 69%
Box 6 2 246 1%
Box 7 6 165 3%
Box 8 24 284 8%
Box 9 56 265 21%
Box 10 116 456 25%
Box 11 23 480 4%
Box 12 179 615 29%
Box 13 17 359 4%
Box 14 104 510 20%
Box 15 86 814 10%
Box 16 12 254 4%
Box 18 66 192 34%
Box 23 1 256 1%
Box 26 103 374 27%
Box 27 350 350 COMPLETE
Box 29 22 122 18%
Box 30 5 193 2%
Box 31 21 302 6%
Box 32 1 158 1%
Box 34 41 398 10%
Box 35 287 439 65%
Box 36 37 418 8%
Box 37 36 487 7%
Box 38 238 424 56%
Box 39 12 282 4%
Box 41 87 572 15%
Box 42 100 910 10%
Box 44 53 201 26%
Box 47 1 466 1%
Box 50 178 198 89%
Box 51 388 939 41%
Box 52 7 609 1%
Box 54 0 205 0%
Box 57 19 420 4%
Box 60 3 183 1%
Box 62 78 564 13%
Box 63 156 345 45%
Box 70 308 347 88%
Box 71 663 663 COMPLETE
Box 72 614 664 92%
Box 73 151 151 COMPLETE
Box 75 3 77 3%
Box 79 199 199 COMPLETE
Box 81 3  488 1%
Box 87 12 604 1%
Box 95 126 147 85%
Box 96 534 539 99%
Box 97 151 295 51%
Box 98 225 499 45%
Box 100 214 429 49%
Box 104 3 502 1%
Box 106 236 581 40%
Box 107 523 542 96%
Box 110 15 671 2%
Box 115 277 307 90%
Box 116 794 865 91%
Box 117 506 853 59%
Box 118 267 880 30%
Box 119 643 990 64%
Box 120 685 685 COMPLETE
Box 121 150 529 28%
Box 122 309 728 42%
Box 123 45 437 10%
Box 124 18 382 4%
Box 135 78 571 13%
Box 139 40 579 6%
Box 141 95 380 25%
Box 149 88 581 15%
Box 150 972 972 COMPLETE
Box 169 221 728 30%
Add MS 35537 733 744 98%
Add MS 35538 824 858 96%
Add MS 35539 883 947 93%
Add MS 35540 947 1012 93%
Add MS 35541 993 1258 78%
Add MS 35547 34 701 4%
Add MS 35549 24 364 6%
Add MS 35550 90 637 14%
Overall 18,355 41,372 44%

Progress Update – 6 to 12 May 2017

By Louise Seaward, on 12 May 2017

Another week, another strong batch of transcriptions submitted by our transcribers. Thank you to everyone who has contributed to the site recently!

18,271 manuscript pages have now been transcribed or partially-transcribed, which is an increase of 47 on last week’s total.  Of these transcripts, 17,276 (94%) have been checked and approved by TB staff.

The more detailed progress chart is as follows:

Box No. of manuscripts worked on No. of manuscripts in box Completion
Box 1 769 794 96%
Box 2 729 753 96%
Box 4 45 694 6%
Box 5 201 290 69%
Box 7 6 165 3%
Box 8 24 284 8%
Box 9 56 265 21%
Box 10 116 456 25%
Box 11 12 480 2%
Box 12 178 615 28%
Box 13 13 359 3%
Box 14 76 510 14%
Box 15 86 814 10%
Box 16 12 254 4%
Box 18 66 192 34%
Box 23 1 256 1%
Box 26 81 374 21%
Box 27 350 350 COMPLETE
Box 29 22 122 18%
Box 30 5 193 2%
Box 31 21 302 6%
Box 32 1 158 1%
Box 34 41 398 10%
Box 35 287 439 65%
Box 36 37 418 8%
Box 37 36 487 7%
Box 38 238 424 56%
Box 39 12 282 4%
Box 41 87 572 15%
Box 42 97 910 10%
Box 44 53 201 26%
Box 47 1 466 1%
Box 50 178 198 89%
Box 51 387 939 41%
Box 52 7 609 1%
Box 54 0 205 0%
Box 57 19 420 4%
Box 60 3 183 1%
Box 62 78 564 13%
Box 63 156 345 45%
Box 70 308 347 88%
Box 71 663 663 COMPLETE
Box 72 614 664 92%
Box 73 151 151 COMPLETE
Box 75 3 77 3%
Box 79 199 199 COMPLETE
Box 81 3  488 1%
Box 87 12 604 1%
Box 95 126 147 85%
Box 96 534 539 99%
Box 97 151 295 51%
Box 98 225 499 45%
Box 100 214 429 49%
Box 104 3 502 1%
Box 106 236 581 40%
Box 107 523 542 96%
Box 110 15 671 2%
Box 115 277 307 90%
Box 116 794 865 91%
Box 117 506 853 59%
Box 118 267 880 30%
Box 119 643 990 64%
Box 120 685 685 COMPLETE
Box 121 150 529 28%
Box 122 309 728 42%
Box 123 45 437 10%
Box 124 18 382 4%
Box 135 69 571 12%
Box 139 40 579 6%
Box 141 95 380 25%
Box 149 88 581 15%
Box 150 972 972 COMPLETE
Box 169 218 728 29%
Add MS 35537 733 744 98%
Add MS 35538 824 858 96%
Add MS 35539 883 947 93%
Add MS 35540 947 1012 93%
Add MS 35541 993 1258 78%
Add MS 35547 34 701 4%
Add MS 35549 24 364 6%
Add MS 35550 90 637 14%
Overall 18,271 41,126 44%

Progress Update – 29 April to 5 May 2017

By Louise Seaward, on 8 May 2017

Rise and shine, it’s Monday! We’re here to give you an update on the transcription that has been done over the past week.  The transcribers are continuing at a good pace and we need to thank them for all their hard work.

18,224 manuscript pages have now been transcribed or partially-transcribed, which is an increase of 42 on last week’s total.  Of these transcripts, 17,222 (94%) have been checked and approved by TB staff.

The more detailed progress chart is as follows:

Box No. of manuscripts worked on No. of manuscripts in box Completion
Box 1 769 794 96%
Box 2 729 753 96%
Box 4 45 694 6%
Box 5 201 290 69%
Box 7 6 165 3%
Box 8 24 284 8%
Box 9 50 265 18%
Box 10 116 456 25%
Box 11 12 480 2%
Box 12 178 615 28%
Box 13 6 359 1%
Box 14 63 510 12%
Box 15 86 814 10%
Box 16 12 254 4%
Box 18 66 192 34%
Box 23 1 256 1%
Box 26 67 374 17%
Box 27 350 350 COMPLETE
Box 29 22 122 18%
Box 30 5 193 2%
Box 31 21 302 6%
Box 32 1 158 1%
Box 34 41 398 10%
Box 35 287 439 65%
Box 36 37 418 8%
Box 37 36 487 7%
Box 38 238 424 56%
Box 39 12 282 4%
Box 41 87 572 15%
Box 42 96 910 10%
Box 44 53 201 26%
Box 47 1 466 1%
Box 50 178 198 89%
Box 51 387 939 41%
Box 52 7 609 1%
Box 54 0 205 0%
Box 57 19 420 4%
Box 60 3 183 1%
Box 62 78 564 13%
Box 63 156 345 45%
Box 70 308 347 88%
Box 71 663 663 COMPLETE
Box 72 614 664 92%
Box 73 151 151 COMPLETE
Box 75 3 77 3%
Box 79 199 199 COMPLETE
Box 81 3  488 1%
Box 87 12 604 1%
Box 95 126 147 85%
Box 96 534 539 99%
Box 97 151 295 51%
Box 98 225 499 45%
Box 100 214 429 49%
Box 104 3 502 1%
Box 106 236 581 40%
Box 107 523 542 96%
Box 110 15 671 2%
Box 115 277 307 90%
Box 116 794 865 91%
Box 117 506 853 59%
Box 118 267 880 30%
Box 119 643 990 64%
Box 120 685 685 COMPLETE
Box 121 150 529 28%
Box 122 309 728 42%
Box 123 45 437 10%
Box 124 18 382 4%
Box 135 64 571 11%
Box 139 40 579 6%
Box 141 95 380 25%
Box 149 88 581 15%
Box 150 972 972 COMPLETE
Box 169 217 728 29%
Add MS 35537 733 744 98%
Add MS 35538 824 858 96%
Add MS 35539 883 947 93%
Add MS 35540 947 1012 93%
Add MS 35541 993 1258 78%
Add MS 35547 34 701 4%
Add MS 35549 24 364 6%
Add MS 35550 90 637 14%
Overall 18,224 41,126 44%