Crowdsourcing based curation and user engagement in digital library design

A historical perspective on the development and success of Trove, including the original idea concept, the user centric design principles and social engagement with thousands of volunteers. The Trove service which is now ten years old is used by millions of Australians with the digitised Australian newspapers being the most popular resource. Rose Holley, Special Collections Curator at UNSW Canberra discusses the findings of her research into crowdsourcing based curation. Using the digitised historic Australian Newspapers as an example, she looks at how the functionality and interface was developed in close relationship with the users, and how this led on to text correction of newspaper articles. It is nearly ten years since this pioneering project began and the motivations and achievements of the 50,000 volunteers are examined over this time. She questions how successfully the goal of improving text quality and therefore search has been achieved. She proposes that if a similar project was begun now then artificial intelligence software would be used such as OverProof post OCR correction tool to improve the quality of the text. OverProof has been trained on the manual corrections of the Australian newspaper corpus and trials demonstrate it is able to dramatically improve the quality of the corpus. Volunteer text correction could still continue afterwards for difficult text but the software would do the main donkey work, allowing users to have a better quality search