From the periphery, I have always been interested in crowdsourcing and have often thought that it would be an interesting way to extend the www.bombsight.org apps that I developed back in 2012,.ie would it be useful for the public to engage in the crowd collection, co-curation of memories, photographs and documents to add to the formal archival resources. Furthermore, here at the CVCE we have been considering if it would be suitable to explore whether crowd-sourcing would be a helpful way of transcribing hand written and other documents, from our Pierre Werner publications or even for error detection and validation of automatic OCR and entity extraction. In our histoGraph tool, the notion of expert crowds has been implemented to annotate and validate automatic face recognition of European Figures from face-recognition algorithms that processed our picture collection.
Therefore, I was interested to find out more about crowd-sourcing projects and research in the Humanities. At the beginning of September I was lucky return to London to attend the Crowdsourcing for the Humanities in the 21st Century event, organised by Kings College London.
The theme of the 2 days of presentation and discussions was “Citizen Humanities Comes of Age: Crowdsourcing for the Humanities in the 21st Century”. The event was made up of a mix of interdisciplinary researchers from diverse backgrounds such as Archaeology, Ecology, Archival Research, Social Sciences and Humanities, which made for fascinating discussion.
For me some of the points of interest collated from discussions and speakers include:
- There is a need for more research around the ethical considerations of crowd-sourcing activities. Is it possible to recognise the skills acquired and time contributed to such humanities projects?
- Crowds alone do not sustain projects, it is essential to build a community from the bottom up (this reminds me of the very successful Open Street Map community);
- Projects should not just be about technology and data capture – there needs to be greater emphasis on the research questions and managing social experiences.
- There is an increasing need to nuture and prioritise human factors for improving likelihood of success.
- Is it possible to develop crowd-sourcing pipelines that mix automation, human-intervention and machine learning to further develop crowd-sourcing activities.
- Project design should enable the framing of dynamic research questions that emerge as projects evolve.
- All projects have difficulty in overcoming the long-tail where the highest volume of contributions is made by a small proportion of all the contributors.
- Successful projects should consider integrating a task ecosystem that provide a mix of different tasks taking different lengths of time to complete- appealing to non-experts and experts depending of the type of task and so eliciting different types of contributions.
- Lots of projects fail but rarely are there discussions of why or how they fail (for me I think lessons learned from failure can be as interesting as those that succeed)