Thread by @natlibscot

During lockdown, Library staff have been improving the quality of transcriptions of our collection of 3,000 digitised Scottish Chapbooks using the @wikisource platform.

#NLSdigitised #NLSData

Wikisource is an online library of out-of-copyright, digitised books. It’s part of a wider family of free, open knowledge project run by @wikimediauk; @Wikipedia is its more famous sibling.

More info about Wikisource > en.wikisource.org/wiki/Wikisourc…

More info about Wikimedia >
wikimedia.org.uk/about/

Using Wikisource is helping to make our digitised books easier to navigate and access through keyword search.

After we digitise a book, we run it through software to automatically creates a transcription by reading the image file and converting it into keyword-searchable text. This process is called OCR (Optical Character Recognition).

Unfortunately, OCR quality is not always perfect, meaning the automated transcriptions tend to contain errors, such as those contained in the image below

We’ve been using Wikisource’s OCR (Optical Character Recognition) tool to improve the quality of these transcriptions so that it is 100% accurate.

Wikisource is free to use, has an intuitive transcription correction interface and contains an in-built quality control function. And by uploading our books to the platform we are increasing the reach of our collections (many books on Wikisource are viewed 100s of times per day).

Our workflow:

1. Books bulk uploaded to Wikisource via @WikiCommons
2. Book transcribed using OCR software
3. Transcription proofread
4. Proofread text validated & published on Wikisource
5. Transcription extracted & reuploaded to our Digital Gallery.

en.wikisource.org/wiki/Wikisourc…

Since late March 2020, 69 Library staff members have taken part, working collaboratively to complete 646 books and proofread almost 12,000 pages between them. At one stage our staff were contributing more to Wikisource than all other editors combined.

Once all 3,000 books are complete, they will also be loaded into our Data Foundry as a dataset of fully corrected OCR transcriptions. View the Data Foundry at data.nls.uk

#NLSData

Through this project we've forged links with the wider Wikimedia community. This included an excellent webinar from @lirazelf & @emcandre, who presented to 30+ Library staff about the value of contributing to the platform, and who trained us in the mystic art of Wikisourcery 😉👍

We’ve also had great support from experienced Wikisourcerors , with particular mention to Beeswaxcandle, EncycloPetey, Xover and @billinghurstwik for their input. ShakespeareFan00 was even moved to write a poem about our work 😊en.wikisource.org/wiki/User:Shak…

Our digitisation manager @GWillshaw will be talking more about the project at the @wikimediauk AGM on 18 July. Sign up here: eventbrite.co.uk/e/wikimedia-uk…

And if you’d like to know more about our Chapbooks, read this short blog post from curator Anette Hagan about the collection blog.europeana.eu/2019/08/chapbo…

Thanks for reading about our Wikisource project 👍