POSTPONED - Using Machine Learning to Automatically Transcribe British Library Collections

The British Library, the national library of the United Kingdom, is estimated to hold between 170 to 200 million items, making it the largest library in the world by number of items catalogued. While many of these are sound and music recordings, videos, prints, drawings and other formats, most of the Library’s items are texts – whether printed books, handwritten manuscripts, journals, newspapers and more.

The Library’s key aim is to “make our intellectual heritage accessible to everyone, for research, inspiration and enjoyment.” Access can be physical or digital, and now we can make use of emerging digital technologies to make our collections even more accessible. Machine Learning and Pattern Recognition can help us convert images of text into machine-readable text. While Optical Character Recognition (OCR) has been around for a while, there are some ground-breaking developments in Handwritten Text Recognition (HTR) in recent years. 

This talk will introduce OCR and HTR, and discuss their importance in the context of library collections. It will also pose some of the challenges to automate text recognition in different collection types and different languages – especially handwritten materials and non-Western languages – and how we try to tackle those challenges.

All are welcome and there will be drinks and discussion after the presentations. Attendance is free but we kindly ask that you register for the event. 

This event is organised by UCLDH, which is part of the UCL Institute of Advanced Studies.