Adapting EATS for Crowdsourcing: Register Medicorum Medii Aevi

Adapting EATS for Crowdsourcing: Register Medicorum Medii Aevi
Viglianti, Raffaele, King's College London, United Kingdom, raffaele.viglianti@kcl.ac.uk
The Register Medicorum Medii Aevi (RMMA) is an active pilot project at King's College London that seeks to lay the foundations for an interdisciplinary online register of doctors in the middle ages. It is common for similar kinds of historical online resources to prepare a database out of a large collection of scholarly findings. RMMA, instead, aims to create a growing database which is populated over time by various scholars and students to make their findings accessible on the web. In order to pursue this task, the prototype for this pilot must allow data insertion from a wider group than the project partners alone; therefore several crowdsourcing principles are applied and extensive user testing is being undertaken. This poster will show solutions to the main challenges encountered in the design of the database and in adapting EATS, a Django application for authority records, for wider online use.

Overview
Researchers from different areas of medieval studies are likely to encounter information in primary sources about medici, the physicians of the time. Information about medici and their practices was recorded in a variety of documents; they would often travel across cultures, which can serve as indication of their high value. This has produced many sources from different cultures and in different languages. Because different areas of study often gravitate around geographical, cultural and linguistic divisions, it is challenging to investigate the movements of people and knowledge across these barriers. RMMA explores how digital publication can help in exchanging reliable information between researchers in such areas and currently involves experts on Anglo-Saxon, Arabic, Armenian, Greek, Jewish, Latin materials that provide information about doctors, and medical knowledge in the medieval period.

To address this issue, the project is developing a web application that will allow to (1) insert information about medici from primary and secondary sources; (2) handle different opinions about the same individuals; and (3) connect to prosopographical projects that deal with a similar historical time-frame and with historical geographical resources. RMMA is looking in particular at prosopographical projects developed at the Centre for Computing in the Humanities at King’s College London, including the Prosopography of the Byzantine Empire and Prosopography of the Byzantine World <http://www.pbw.kcl.ac.uk/>. For historical locations, we are working towards integrating a query system to Pleiades when entering new locations <http://pleiades.stoa.org/>.

Technologies Involved
The application is built using Django,http://www.djangoproject.com/ a Python open source web application framework aimed at simplifying development through a reusable set of common libraries and components. Developers can code their own application to be re-used within other Django frameworks, which is a factor that contributes towards the web framework's success. EATS (Entity Authority Tool Set) is a Django application developed by Jamie Norrish at the New Zealand Electronic Text Centre (NZETC) and "is now used to manage the 30,000+ entities represented in the NZETC online collection of significant New Zealand and Pacific Island texts" (Norrish 2007). The application targets issues usually faced by libraries, such as ambiguous identifiers based on names and titles. By improving on traditional authority systems where one identifier groups different entries describing the same entity, EATS allows the linking together of different collections and authorities, introduces user management and provides greater scope for disambiguation. RMMA found that EATS's approach could greatly contribute to the objectives of the project's web application, specifically in handling information about the same entities coming from multiple sources.

Most EATS top-level components have been adapted for RMMA's purposes: Entities are database tables that must be associated to one or more Authorities. Authorities may then make Property Assertions that associate Entities to specific default or user-defined properties. In RMMA, Entities are medici, patients, employers, locations, institutions, etc., while Authorities are the contributors of the data, such as authors of narrative secondary sources or contributors inserting information anew from a primary source (note that in this case the authority remains the contributor and not the primary source; in theory, different contributors can enter different information from the same primary source, intentionally leaving room for debate). Property Assertions in RMMA reflect statements such as: Authority contributor states that Entity medicus has name name, or similar. Finally, users of the system are able to insert data for one or more authorities, which is particularly useful if one user is entering information from a number of secondary sources. When users enter data discovered by themselves, however, they are acting like authorities. Even in this case, for the purpose of the system, the dichotomy between user and authority must be maintained: the scholar will insert data as a certain user controlling the authority that represents him or herself.

Fig 1. This graph shows the role of users in controlling authorities and different authorities declaring statements (property assertions) about the same entity.

Full Size Image

Adapting EATS Interface for a Wider Audience
The default EATS interface for data entry is designed for completeness and may require some training for the inexperienced users. After all, this interface is likely to be used by personnel for populating the database and most of the design effort is better spent on the querying interface that end users will see. In the case of RMMA the project is vastly adapting the interface, as data entry will be an essential part of the end user experience. The project has been looking at other initiatives in the Digital Humanities and Digital Library fields that are adopting crowdsourcing for collecting data.See in particular: Transcribe Bentham, a crowsourcing project hosted at the Centre for Digital Humanities, University College London <http://www.ucl.ac.uk/transcribe-bentham>; the image collection of the Victoria & Albert Museum <http://collections.vam.ac.uk/crowdsourcing>; and other projects discussed in Holley 2010. Crowdsourcing consists of outsourcing specific tasks to be performed by volunteers via a web application. The term is a neologism coined in 2006 in an article on Wired, in which the author claims that the current pervasiveness of technology has reduced the gap between professionals and amateurs. Rose Holley (2010) argues that crowdsourcing could be of great value for Digital Libraries, especially for accomplishing goals that would require a large number of staff, time and resources. In order to attract contributions from volunteers it is necessary to follow certain principles when designing the web application that they will use. Holley outlines some "tips for crowdsourcing" out of her personal experience and by analysing successful projects.

RMMA is not strictly a crowdsourcing project, as submissions will be reviewed by a board before being published. The reason for this is twofold: on one hand it is desirable to control what gets into the database, though successful crowdsourcing initiatives usually found volunteers good at policing each other and keeping away malicious users; on the other hand, RMMA attempts to promote the use of the database as a publishing environment for scholars. The presence of a board, therefore, offers a model closer to peer-reviewed publishing, which is not a novel approach, as some academic projects have already been experimenting along similar lines.An exemplary project is Suda on Line: Byzantine Lexicography (SOL) <http://www.stoa.org/sol>, which involves more than a hundred scholars for the translation and editing of the Suda, a 10th century CE Byzantine encyclopedia. The large size of Suda (30,000+ entries) has encouraged the involvement of a large number of collaborators to work online. Submissions to the database are made immediately available to the public, but SOL implements a colour code that differentiate just submitted entries and entries that have been reviewed and approved by a the editors overtime. This mixed system simplifies submission whilst allowing a control of the quality of the publication.

Despite these plans, which move RMMA away from the definition of crowdsourcing, many of the principles outlined by Holley are of great importance. Specifically, she suggests having a clear and ambitious goal, acknowledging contributors, reporting on the progress of the resource and making the application easy, reliable, quick and intuitive. Current efforts focus on improving the application by simplifying the interface and making clear paths for data entry depending on the nature of information that the users intend to store. To insure that the interface design is effective, several stages of user testing are planned.As recommended by many studies on User Centred Design, after Norman 1988. RMMA is organising a number of workshops in Europe and in the United States as part of the pilot project. This is intended to reach the desired audience and at the same time have the user test the resource and collect feedback during the following few weeks.

The Future
RMMA hopes to use its web application to involve middle ages medical researchers from different fields separated by specific geographical and linguistic studies. The database aims to be both a growing collection compiled by interested participants and a reliable publication of information about relevant individuals from the middle age worlds. It is therefore necessary to include crowdsourcing principles in the design of the web application to achieve the proposed goals. While not a stated aim of the project, the possibility of tracing of the movement across cultures of individuals conveying and acquiring medical knowledge would be a desirable and likely outcome of far-reaching historical interest.

References:
Finkel, R. et al. 2007 The Suda On Line, 31 November 2010 (link)

Holley, R. 2010 “Crowdsourcing: How and Why Should Libraries Do It?, ” D-Lib Magazine, 16 3/4

Norman, Donald A. 1988 The design of everyday things, MIT Press Cambridge (MA, USA)

Norrish, J. 2007 “EATS: an Entity Authority Tool Set, ” Australia New Zealand Digital Encyclopedias Group Meeting, Sydney, Australia, 7-8 December

Norrish, J. and Stevenson, A. 2008 “Topic Maps and Entity Authority Records: an Effective Cyber Infrastructure for Digital Humanities, ” Digital Humanities 2008, Oulu Finland, 25-29 June

Terras, M. 2010 “Crowdsourcing Cultural Heritage: UCL’s Transcribe Bentham Project, ” Seeing Is Believing: New Technologies For Cultural Heritage, International Society for Knowledge Organization. University College London, London, 9 June 2010