Digital humanities projects and standards: Let’s get this conversation started!

2.1.Preserve the Baltimore uprising

The Preserve the Baltimore Uprising Archive Project is a digital archive that was designed originally to document first-hand experiences of the protests and calls for action in the Baltimore area following the death of Freddie Gray in 2015. In response to the protests for Black Lives Matter in 2020, the Preserve the Baltimore Uprising project has decided to expand its collection to include content for protests related to murders of Ahmaud Arbery, Breonna Taylor, George Floyd, and others [4]. This project excels at providing space for first-hand and official accounts of protest to exist side by side in the same web space. It also offers an important pedagogical space, to help train those interested in methods and theories of recording oral history [5].

Metadata created for Preserve the Baltimore Uprising reflects a grassroots approach to collecting and preserving content. Contributors submitting content create descriptive metadata for images, interviews, or written content (the content is vetted based upon a collection policy that focuses on relevancy to events of protest and unrest) [6]. The metadata is structured using the Dublin Core schema, which offers a great deal of flexibility for users to choose the specific fields that they need to accommodate a project. Dublin Core is flexible enough and accessible enough that even inexperienced metadata creators can contribute meaningful and important contextual information to describe and catalog content. The relatively open metadata creation policy is both a strength and a weakness of this platform. The collection is small enough that end users can find specific content inside of the platform without using the search box, but overall, the searching infrastructure under Search Items is not designed for discovery searching (known-item searching is easier, especially if the user knows the ID#). Unfortunately, no explanation is offered to help users effectively use the search under Search Items menu. The search menu is built around the underlying metadata element sets chosen from Dublin Core, Item Metadata, and Zotero, but it feels overlaid rather than growing organically out of user experience [7]. The two best ways to find content are to use the Browse by Tag menu or the Browse Collections menu. A quick look at the tags used in the Browse Tags menu reveals that these labels are likely a folksonomy rather than a controlled vocabulary imported from existing practices and standards developed previously [8]. The collections menu is organized by individual contributors who contributed the content [9].

Fig. 1.

Francis Bacon’s ISNI record, complete with names of people who are known or supposed to have a connection with him.

Fig. 1 (Continued).

Fig. 2.

Francis Bacon’s Library of Congress Name Authority record, complete with links to authoritative forms of his name in other schemas.

Fig. 2 (Continued).

2.2.Six degrees of Francis Bacon

The Six Degrees of Francis Bacon is directed specifically to an academic audience, but open to community contributions. It is “a digital reconstruction of the early modern social network that scholars and students from all over the world can collaboratively expand, revise, curate, and critique” [10]. The graph that represents the entire network reflects who was connected to whom and what types of people were connected (e.g. lawyers, composers, philosophers) in early modern Britain.

From a user perspective, this platform is intended for historians who study early modern Britain. From a metadata perspective, however, this network is an excellent example of how choosing well-developed sets of metadata can make a project interoperable with others. The directors of the project created crosswalks between the names recorded in the Francis Bacon database and authoritative name sets in the Wikidata database [11]. The name authorities imported included the Virtual Integrated Authority File (VIAF), the German Integrated Authority File (GND), the International Standard Name Identifier file (ISNI), and the Library of Congress Name Authority file (LCNAF). The project partners have also linked the Oxford Dictionary of National Biography [12].

These discrete data points from controlled vocabularlies, enhanced with human intervention, work together in the Six Degrees of Francis Bacon to create a visual representation of the social world active in early modern England. The controlled vocabularies (e.g., VIAF, GND, ISNI, LCNAF) make the project interoperable with other projects and collections that use these same authority files. In addition to the officially recorded version of a person’s name, the controlled vocabularies provide other important information about Francis Bacon's birth and death dates, occupations, aliases and alternate name spellings (see for example Francis Bacon’s ISNI (Fig. 1) and LC Name Authority Records (Fig. 2)). The bibliographical notes in the ISNI and LCNAF are then complimented by the biographical entries in the Oxford Dictionary of National Biography, thus fleshing out a fuller narrative of a person’s life. Dates, occupations, and biographical information provide the data points to buildout the timelines and group networks. All of this metadata, once ingested, can create networks automatically with the help of a Python script. Humans further enhance it based on expertise and supplemental knowledge.

Fig. 3.

South Sudan Diversity Cam Image with controlled vocabulary of animals listed on the right side of the screen.

Fig. 4.

Image of Fishing at the Sunnyside Dam.

Fig. 4 (Continued).

Washington State University Metadata for Image of Fishing at the Sunnyside Dam.

Fig. 4 (Continued).

Yakama metadata, with Description, Cultural Narrative, and Traditional Knowledge for the Image Fishing at the Sunnyside Dam.

Fig. 4 (Continued).

Yakama metadata, with Description, Cultural Narrative, and Traditional Knowledge for the Image Fishing at the Sunnyside Dam.

2.3.Zooniverse

All projects discussed in this paper have a crowd sourcing component that shares the labor of creating content and metadata. Zooniverse is, however, the project that is most active with recruiting and training volunteers to participate in citizen science projects: “At the Zooniverse, anyone can be a researcher” [13]. Most of the projects include work whereby participants add or clarify information depicted in an image or video. A particular strength of this project, in addition to the crowd-sourced metadata created in individual projects, consists of articles documenting the challenges and effective execution of citizen science projects [14].

So what about the metadata for projects indexed in Zooniverse? Each project has produced or is producing a plethora of associated descriptive metadata, to help researchers identify and study large content-rich, non-machine readable data points, including images of space, illustrated manuscripts, letters, and other handwritten documents. For many projects, the project directors provide the metadata to attach to an image, text, or video, which helps to control the vocabulary and descriptive terms internal to the project. For example, in a project measuring animal diversity in the South Sudan, volunteers select animals from a defined list based on what they see in the image. From a metadata creation perspective, this helps the research directors use a particular spelling of each animal and also defines which animals to observe in the camera (South Sudan DiversityCam, Fig. 3). Content description of these types of materials is cost prohibitive for most institutions or research groups, which is why Zooniverse offers such a great opportunity to help researchers conduct otherwise difficult visual and textual analysis while giving volunteers a meaningful way to contribute to a project or study a topic about which they are passionate [15].

While the descriptive enhancement provided through the Zooniverse project model is robust, the submission policies for the platform do not require specific sets of metadata to be deposited into the platform. Some projects come with pieces of metadata that are required in many publishing platforms, including date of creation, author, or DOI/persistent URLs; other projects do not and may also lack distinct subjects, topics, names or other pieces of metadata associated with the objects before volunteers work with them. The metadata designed for each project grows out of the project needs rather than being drawn from standardized vocabulary. From a large-scale indexing standpoint, this means that the best way to facilitate connections between projects in the Zooniverse platform is to rely on the hashtags created by the tagging volunteers or by the discipline tags assigned by the platform managers (the current tags are: arts, biology, climate, history, language, literature, medicine, nature, physics, social science and space).

2.4.Mukurtu CMS

Like Zooniverse, the Mukurtu Content Management System (CMS) is a software tool and not a specific project, but at least two major projects use this CMS to digitally archive sacred and historically-significant tribal collections [16]. The history of Mukurtu and the metadata management provided within the platform are bound together. Mukurtu was conceived as a tool to “manage, circulate, and narrate the Warumungu Aboriginal community’s digital materials using their own cultural protocols” [17]. The materials retained digitally in the original Mukurtu platform, the Mukurtu Wumpurrarni-kari Archive, are culturally-sensitive objects that were originally collected and archived without following Warumungu tribal protocols. Mukurtu is helping to fix the problematic archival practices of the local, state, and national archives that appropriated these materials from the Warumungu people. The second major installation of Mukurtu was the Plateau Peoples’ CMS, which “sought to extend the alpha version of Mukurtu to incorporate existing library digital collections (with Dublin Core metadata), in a web-based platform including multiple tribes across several states who share common histories, but also unique tribal vales, languages, and collections” [17].

The metadata fields for the Warumungu and Plateau People’s instances of Mukurtu are tribe-specific. Among the most important features of Mukurtu’s metadata schemas are the self-defined fields, that give communities the power to categorize and name in culturally-appropriate ways, and flexible rights access, which allows platform designers to control rights access to archival content based upon a user’s relationship to the content owners [18]. The customization features of Mukurtu are one of its strengths and force users to engage with collections in a particular cultural, social, and historical context driven by content owners. It supports what founder Kimberley Christen calls “digital heritage stewardship”, which shifts management of collections from “collecting to stewarding [undoing] the colonial model and [inscribing] a relational model built on obligations that one has to care for, maintain, and preserve a variety of belongings” [19]. In other words, Mukurtu makes it possible for collections to be curated and maintained in a collaborative way. In the Plateau People’s Portal, for example, images can have multiple sets of metadata that allows a rich and layered history of a given object to come through (e.g., see Fig. 4a an image of fishing at the Sunnyside dam (Fig. 4b with Washington State University Metadata and Fig. 4c with Yakama Metadata)).