Managing organisational impact

The Zooniverse guide to ‘building a great project’ begins ‘[kjnow that you are making a commitment!' ('Part I: Building a Great Project’, n.d.). Crowdsourcing projects require ongoing attention from staff and assessing whether you can provide resources for the life of the project is an important step in assessing the feasibility of a project. Staff7can be supported by volunteers for some tasks, such as answering questions from other participants, but they must also have time to report on progress to internal and external stakeholders, and prepare newsletters and social media updates for outreach and marketing purposes.21’ Staff might also need support in gaining new skills such as community management or workflow integration.

Crowdsourcing projects can have an impact on the workload and outputs of departments across the organisation. For example, they can lead to increased attention to collections, and requests for new or reprioritised digitisation to keep items flowing into the platform. If your project is to generate metadata, annotations or other information about collection items, talking to the teams that manage the systems that store information about collections is vital. They can specify import formats and help you determine what information will be most useful to collect to improve catalogue or discovery systems. If information collected does not fit into existing interfaces (for example, your collections management system has no capacity to store user-generated content), where will it be kept? Ensure technical documentation is shared with relevant staff even if the platform is developed externally. The work of preparing material for ingest into the platform, and of reviewing and packaging task results for ingest into internal systems should also be included in overall resource plans.

Finally, an internal communications strategy, however informal, will help the rest of the organisation feel involved in the success of the project. You can share updates via internal presentations and emails, and invite staff to test interfaces, brainstorm ideas for outreach methods to reach potential participants, and plan publicity in physical venues.

Choosing source collections

Crowdsourcing relies on the availability of digitised collections. Digitisation can be expensive and timeconsuming, so get estimates for delivery dates before building milestones around new digitisation. Source collections may also be determined by the goals of the project—research projects on a particular topic may choose items in a range of formats, while a project aimed at increasing discoverability might work through one format at a time.

Some content has a wider immediate appeal, and consequently makes the work of recruiting participants easier. Lascarides and Vershbow said 'it is much easier to get patrons excited about participation in a project if they are already excited about the source material’ when describing the choice of material for what become fWiuti on the Menu (2014). DIY History selects handwritten, historically significant,‘interesting’ and extensive materials (DiMeo, 2014).They also note a preference for material is that is ‘old enough' to avoid copyright and privacy issues. While collections that appeal to both casual viewers and scholars make attracting interest and justifying participation much easier, it is possible to create compelling stories about more obscure collections or to invite specialist communities to become involved.

Planning workflows and data re-use

The source material—text, digitised images, audio-video, etc.—and goals of the project determine the types of tasks that will be crowdsourced. Planning the workflows necessary to make data usable is part of the process of assessing the feasibility of a project: there is no point asking people to help create data or knowledge that you cannot use as intended. Creating a workflow plan is part of managing the organisational impact and integration of a project, and should ensure that you can move digitised source material into your crowdsourcing platform, then move validated data to the system (which might be a collections management system, web publishing system etc.) in which it can be used. Collections management staff can advise on the most useful data for discoverability or help work out how to publish research datasets. Any changes required are likely to take time to implement so begin conversations with relevant departments as early as possible.

Source items might need pre-processing before they are presented in tasks. For example, some projects categorise manuscript items by how easy or hard they are to transcribe. Some pre-processing tasks can be built into the task ecosystem, such as Fossil Finder,2 which asked participants whether a photo was ‘good enough to study’, instructing them,‘If it is too blurry, dark, noisy, or bushy then bin it!’. Workflow also includes task validation and quality control processes, although these are usually built into the crowdsourcing platform. Data might also need post-processing to convert it into formats suitable for ingest and sharing with project contributors.28

Workflows should be tested as early as possible to allow time to manage any logistical, technical, legal or institutional issues that arise. This behind-the-scenes work ensures that new items can be easily added to the platform, and that data created is put to work as soon as possible, helping demonstrate the value of volunteer contributions to all.

Workflow systems should be designed for modularity to allow for changes in other platforms over time. Collections management systems can be refreshed, new tasks with different export formats devised - or you may start to integrate machine learning processes to create human computation systems. Finally, in order to re-use content created by volunteers, you should ensure that you have put in place terms and conditions that give you the right to use the data.

< Prev   CONTENTS   Source   Next >