Due to its complexity and necessarily broad range of required skills, the development of cyberinfrastructure often requires collaboration between humanists and non-humanists. First and foremost, scholars of the humanities must lead the development of their own cyberinfrastructure, despite the fact that tool development is not yet rewarded on par with more traditional scholarly outputs [Schreibman and Hanlon 2010]. They alone can accomplish the intellectual task of modeling scholarship for new digital environments [Drucker 2009]. They alone can judge its usefulness and assure that the technology will serve their needs [Hunt et al. 2011]. However, few individual humanities experts (let alone any expert) will have the range of expertise required to build cyberinfrastructure, so librarians, archivists, programmers, and computer scientists will bring essential complementary skills to the task [Borgman 2009]. Technology developers and humanities researchers working alongside each other are likely to push both the technical and the intellectual boundaries farther, yielding greater innovation [Collins et al. 2012]. Casting the net even wider, the ACLS report, which lists collaboration as a necessary characteristic of cyberinfrastructure, calls for “cooperation with librarians, curators, and archivists; the involvement of experts in the sciences, law, business, and entertainment; and active participation from and endorsement by the general public”  [ACLS 2006]. However, it should be noted that, while necessary, collaboration is itself a challenge that requires careful resolution of methodological differences and regular communication about each collaborators’ perspective [Chuk et al. 2012].

In the case of SNAC, its emphasis on social networks is conducive to attracting collaborative partners. The study of social networks has been identified as a cross-cutting agenda where the digital scholarship of humanists, technology researchers, and others converge in a way that encourages collaboration and, in particular, brings humanists to a seat at the table where scholarly infrastructure decisions are being made [Borgman 2009]. The social network is a topic that lends itself to interesting collaborations among many different disciplines with a scope not so narrow as to constrain research nor so expansive as to be meaningless [Friedlander 2009].

The Institute for Advanced Technology in the Humanities (IATH), University of Virginia, leads the team developing SNAC. IATH is known for collaborative research and technology development with the practical needs of humanities scholars as its primary motivation. The list of other participating institutions and individuals on SNAC’s project team and advisory board suggests a wide range of disciplinary expertise is contributing to the project, including historians, English scholars, librarians, archivists, information technology developers, information scientists, computer scientists, and more [IATH n.d. c]. The breadth of knowledge required by the development of SNAC means no single discipline has all the answers to the project’s research questions. Archivists, librarians, and information scientists are needed for their expertise in existing infrastructure surrounding primary resources, experience in information organization, and their knowledge of new data standards. Computer scientists and developers contribute by solving and implementing solutions for the myriad of technical challenges encountered during the course of SNAC’s development. Historians and English scholars contribute their knowledge of humanities research methods and SNAC’s user community, helping shape the public interface to the data and ensuring the new tools are designed to be of practical value to scholars. For the sake of brevity, these examples are probably oversimplified. The full picture of collaboration on SNAC is more complicated, especially since its participants often wear multiple hats and it is hard to encapsulate all the multidisciplinary challenges faced by SNAC in a few short sentences. However, the team does appear to have a mix of expertise well suited to the task of developing SNAC.

Given the great need for teams of collaborators, it should be evident that the scale of cyberinfrastructure research is large, the goals are many and multi-faceted, and the aim is to serve a large audience. However, attempting to do too much, to make a technology useful for everyone in the humanities, risks the possibility of making it useful for nobody, because a tool with a very broad audience in mind is unlikely to meet the requirements of specialized fields [Hunt et al. 2011]. At the other end of the scale, cyberinfrastructure of the past decade has led humanities researchers to create smaller, isolated projects, the contents of which cannot be easily combined for analysis without the development of a more robust infrastructure to link them together [Blackwell and Crane 2009] [Crane et al. 2009]. Creators of a cyberinfrastructure tool should have a specific audience in mind and build features into a tool that serves that audience well, but the tool should also be interoperable with other systems to support some unimagined uses by unanticipated users. A successful cyberinfrastructure of the future will likely consist of multiple systems that exist to support the specialized needs of each discipline, yet support the sharing of data between these systems [Hunt et al. 2011].

SNAC’s focus on social networks and creator description limits its scope by reducing the types of data to be managed and the kinds of functionality designers might build into the tool for analysis. However, it does not restrict the size of the project. SNAC is poised to expand widely within a certain band of information, specifically the description of primary resource creators and the relations between them. SNAC has compiled over 2.6 million EAC-CPF records describing persons, families, and corporate bodies [IATH 2014] since the project began. In future phases of development, nearly 4 million EAC-CPF records may be generated, creating an unprecedented collection of data about the lives, work, and events surrounding historic persons [IATH n.d. b].

With its narrow, but deep, focus on creator description, SNAC intends to do a handful of tasks extremely well and integrate itself into existing infrastructure to offer researchers more functionality. The challenges involved in creating, managing, and providing access to this massive amount of data are a big enough task for one project, so SNAC will rely on others to extend its functionality. For example, SNAC does not intend to collect digitized primary resources or provide direct access to such resources, but rather prefers to leverage existing cyberinfrastructure by providing links to WorldCat (http://www.worldcat.org/) and the online finding aids of archival institutions to bring researchers closer to the resources they seek. Also, SNAC has a specific audience in mind that is large, but certainly not all-inclusive: anyone interested in historical research. SNAC will serve this audience in a few ways: integrated access to distributed primary and secondary resources about people and organizations; access to historical and biographical descriptions; and access to the social networks of people of historical interest [IATH n.d. b]. The emphasis on a small number of functions for a targeted audience built atop a unique dataset of unparalleled size strikes an appropriate balance on the Y-axis between boutique projects and general-purpose applications.

Humanities cyberinfrastructure should be a step forward for research infrastructure otherwise there is no reason to build it. It should offer new solutions to old problems or enable humanists to ask new questions. Therefore, to build such systems there is a need to utilize methods and technologies that go beyond the traditional tools of the humanities. Traditional humanities infrastructure includes intellectual categories such as literary genres and linguistic phenomena, material artifacts like books and maps, buildings such as libraries and book stores, organizations including universities and journals, business models such as subscriptions and memberships, and social practices such as publication and peer review [Crane et al. 2009]. In the digital era, however, existing infrastructure must not be used uncritically as the default model for new infrastructure [Svensson 2011]. New infrastructure must go beyond the capabilities of what already exists, otherwise it is merely reinventing the wheel. Humanists may have something to learn from scientists and engineers who have already defined, designed, and deployed much more in the way of cyberinfrastructure, but this would leave the humanities in the position of building on the technologies constructed for other disciplines [Borgman 2009] and this approach should also not be uncritically adopted [Svensson 2011]. A balance must be struck with new technologies that push the boundaries of scholarly activities, yet remain accessible and meet real needs. Humanities scholars will be unlikely to engage with new technology for its own sake [Collins and Jubb 2012] and their adoption of any new technologies will be highly influenced by existing research practices developed when using traditional infrastructure [Collins et al. 2012].

SNAC pushes the boundaries of traditional infrastructure through its use of digital data. The creator description data that drives the project is a standards-based, internally consistent dataset with high potential for interoperability with other systems and purposes. Data of this sort (and the tools that can use it) provide a desirable characteristic of cyberinfrastructure: that access to the data should be seamless across repositories [ACLS 2006]. Having the potential for data sharing and re-use baked into the initial design of the project is an approach more readily found in scientific research. Also, the rigor with which the data is curated opens the door for innovative approaches to research and the tools that support it. This sort of data-centric approach enables structures and relationships to emerge in the system that would otherwise go unnoticed without a lucky combination of serendipity and painstaking manual research by humans. For example, during the first phase of research and development, the SNAC prototype included an experimental radial graph feature that demonstrated the ability to visualize and explore the relationships between people and organizations represented in the data. See Figure 3.

The radial graph visualization initially puts the name of an individual at the center of a circle of nodes (other names) to which he or she is related in some way. The example in Figure 3 shows “Oppenheimer, J. Robert” as the central node with lines extending out to the names of all the other people and organizations he had been associated with in his lifetime. In addition to the lines drawn between the central node in the network and all the other surrounding nodes, relationships between the surrounding nodes in the circle are also indicated with a line. The visualization enables further exploration by responding to mouse clicks on any node in the graph by redrawing the visualization to include the immediate social network of the selected node. For an amusing example, from the graph in Figure 3, clicking on the node for “Eliot, Thomas Stearns” opens his social network and reveals that Oppenheimer, the father of the atomic bomb, was only two degrees of separation from Groucho Marx[1] (see Figure 4).

While the current phase of SNAC development does not feature the radial graph, the feature is archived at another location (http://socialarchive.iath.virginia.edu/xtf/view?mode=RGraph&docId=oppenheimer-j-robert-1904-1967-cr.xml). It stands as an early experiment in what is possible with SNAC's data, providing a hint of what is to come in future features such as a graph visualization of social-document networks, visualizations of chronologies, and geographic displays of collection locations [IATH n.d. f]. SNAC’s web prototype and its visualization tools like the radial graph are the public interface to (and a product of) a variety of innovative, yet practical research agendas tackled by the SNAC project. The research never sought to build and explore technologies just for the sake of creating new technologies, nor did it adopt any existing technology without a plan to apply its use to the current needs of SNAC’s target user group. SNAC balances the utility of traditional methods that existing infrastructure enable with the exciting, but unproven potential of new technologies. It does so through the use of advanced technology to accelerate and extend traditional methods in new ways. While researchers may have always wanted to search for every occurrence of an individual’s name in the archives, it has previously been impossible due to a lack of funding and time [Collins and Jubb 2012]. However, as some historians have already observed, a year’s worth of work using the current infrastructure combined with a great deal of serendipity and persistence can be replaced by the data in a single SNAC record [IATH n.d. b]. Through this dramatic reduction in time required to conduct research SNAC demonstrates the benefits of a well-designed cyberinfrastructure tool.

To build upon existing infrastructure an understanding of its layered history and the relationships between its various components may be beneficial. For example, when trying to introduce a new public utility or a new mode of transportation to a community, thorough knowledge of the existing layers of infrastructure may reward planners with insights such as knowing the gaps in existing service that a new one may fill, discovering ways to save development effort for a new project by integrating with current infrastructure, and understanding usage patterns that may help predict the best places to initially roll out a new service. Look at the streetscape of a typical city and you’ll see the layers, old and new. Sewers, sidewalks, bike lanes, buses, and telephone poles line the streets. The poles are slung with wires that carry electricity, telephone, cable TV, and the information superhighway. The infrastructure that surrounds us is the result of an iterative process of change and evolution over time. Humanities cyberinfrastructure will be no different.

SNAC is not possible without the many layers of humanities infrastructure that have come before it. SNAC is built on top of a long-standing infrastructure stack that has served the scholarly community in a reciprocal relationship, where it has been shaped and honed by scholarly use and in turn has also shaped how scholars do their research. Any attempt to build cyberinfrastructure ought to include a process of looking back while moving forward. Which facets of prior infrastructure continue to be important today? How can new technology be used to augment and extend those facets to create a new layer of the infrastructure stack? Of course, some may ask, is it even worth building? Certainly there are some who will have doubts about the value of cyberinfrastructure and this points to the importance of the Z-axis of the evaluation criteria. Cyberinfrastructure will not emerge unless it is designed in a way that balances traditional research methods along with the pursuit of advanced technology. One way to design technology that respects the traditions of the past is to look back at the infrastructure stack upon which the new technology is being built and not neglect the scholarly activity it supports. If one can build a technology that eliminates artificial or unnecessary restrictions on scholarly activity, freeing scholars to do what they really want to do — read, write, analyze, produce knowledge, and distribute it — then that technology will be successful [Juola 2008]. What follows is a discussion of several key pieces of the stack beneath SNAC.

Arguably one of the earliest forms of infrastructure supporting information for and about humans is the practice of naming people and their organizations. The abstract appellation is a shorthand way of referencing a whole person in narratives, news, rumors, documents, and data systems like SNAC. Look back at Figures 3 and 4 in this article and notice the primacy of names in SNAC's radial graph. Figure 5 shows a portion of the SNAC prototype's home page where a searchable list of names provides access to data contained in SNAC. The history of naming is long, but is perhaps most significantly related to SNAC in the way personal names reflect the growing complexity of society. For example, the legal and judicial systems of expanding towns in medieval Europe needed to identify individuals clearly in order to attach them to property, to tax and recruit them, and to prosecute them, eventually leading to compulsory naming in the modern era for these reasons [Wilson 1998]. All human organizations now have some bureaucratic interest in names and the official records of many institutions, from churches to governments to corporations and beyond, can end up residing in the next level of the infrastructure stack, the archive.

Archives are institutions that collect the unique records of corporations and the papers of individuals and families, the unselfconscious by-products of corporate bodies conducting business and people living their lives [Pitti 1999]. Archives are part of our culture's memory infrastructure, collecting material of legal and historical value. “Any record of human experience can be a data source to a humanities scholar”  [Borgman 2009]. Historians list just about every conceivable type of material as important to their work, including primary source material such as reports, wills, deeds, diaries, letters, and so on, all materials to be found in the archives [Case 1991]. What tool can a scholar use to find the needles in the haystacks of archives? That would be the next level of the infrastructure stack.

A finding aid is a printed description of all the records left in an archive with a common creator or source. A finding aid contains a description of the creator, functions performed by the creator, and the records generated by the creator through the performance of those functions [Pitti 2006]. The finding aids provide context to archival collections and detail their contents [Daines and Nimer 2012]. Reading through finding aids has been listed by some historians as the most frequently used method for locating primary source material [Dalton and Charnigo 2004]. The archival descriptive principle that organizes material based on its provenance (who created the collection) causes personal and organizational names to be a key access point for finding information in archives. Of course, this creator attribute of a finding aid is not the only name. Deeper sections of the finding aid describing the contents of a collection will also contain a wealth of names. Experienced researchers often know the names of key people or organizations connected to their topics of interest and pursue source material linked to those names to provide context [Duff and Johnson 2002]. Reading finding aids and collecting names found therein is a method for building up a list of leads to new sources. Finding aids have been an integral part of humanities infrastructure, but for a long time the descriptive practice did not conform to expectations of consistency one might expect from such a system. Archivists initially developed specialized local practices for describing their collections, believing their collections to be too unique to employ standardized practices. However, over time, archivists came to realize that although the contents of their collections are unique, the properties, characteristics, and needs of these collections are not [Theimer 2011]. This realization, combined with the energy behind automating digital forms of description, led to the adoption of the structured standard embodied in the next level of the infrastructure stack.

The Encoded Archival Description (EAD) Document Type Definition (DTD) is a standard for encoding archival finding aids using Extensible Markup Language (XML) [Library of Congress 2011 a]. Development of the data standard began in 1993, currently exists in a version released in 2002, and has been adopted by archives to enhance access to their collections by displaying the finding aids online and enabling full text indexing for easier keyword searching [Pitti 1999]. The schema is intended to represent traditional print finding aids in machine-readable form and contains elements to hold the same information one would find in those finding aids, such as the creators, their functions, biographical information, and descriptions of individual records in a creator’s collection [Pitti 2006]. While EAD transformed archival description and brought it into the digital age, the print conventions retained in the presentations of online finding aids has been criticized as inappropriate for the Web and archivists were urged to explore different ways of presenting finding aid content online [Daines and Nimer 2012]. Archivists themselves increasingly recognize that the structure of the traditional finding aid, which is embodied in EAD, is inflexible and inefficient when dealing with complex, interrelated records [Pitti 2004]. The next level up in the infrastructure stack, the aforementioned EAC-CPF, provides the key to revolutionizing archival description and online access. However, before I address that, there’s one more part of the stack that bears mentioning.

Before reaching the top of this mostly archival infrastructure stack a side step to library infrastructure is necessary. An authority record is “a tool used by librarians to establish forms of names (for persons, places, meetings, and organizations), titles, and subjects used on bibliographic records”  [Library of Congress 2011 b]. Of particular interest to SNAC are the collection of authorized names (such as “Twain, Mark, 1835-1910”) and the list of cross-references that lead to the authorized names (such as “Snodgrass, Quintus Curtius, 1835-1910” and “Clemens, Samuel Langhorne, 1835-1910”). For reasons that will be described shortly, SNAC relies on authority files to flesh out the biographical descriptions of people and organizations in its data.

Encoded Archival Content – Corporate bodies, Persons, and Families (EAC-CPF) is the top of the infrastructure stack that supports SNAC. EAC-CPF Schema is a standard for encoding contextual information about persons, corporate bodies, and families related to archival materials using XML [Staatsbibliothek zu Berlin, n.d.]. EAC-CPF solves a problem inherent in traditional finding aids and EAD, that is the difficulty of representing complex records with a single document. Archival records are often of mixed provenance or the records of the same provenance can be dispersed over numerous archives. This situation causes unnecessary redundancy and duplication of effort [Pitti 2004]. EAC-CPF solves this problem by enabling the separation of creator description from record description. Maintaining a unique, centralized creator record not only reduces redundancy and duplication of effort, but also provides an efficient means of linking creators to the functions and activities carried out by them, to the dispersed records they created or to which they are related, and to other related creators [Pitti 2006].

As I mentioned before, one of SNAC’s primary objectives is the creation of a large (and growing) collection of EAC-CPF records that power its prototype access system. As its original National Endowment for the Humanities (NEH) proposal describes, SNAC takes three steps to generate creator descriptions in EAC-CPF [IATH n.d. a]. First, preliminary EAC-CPF records are derived from name and biographical components of EAD. The structured nature of the EAD records enables information to be migrated directly into the EAC-CPF format through automated EXtensible Style Sheet Transformations (XSLT). A less simple part of this first step is the extraction of names from semi-structured and free form text fields that require natural language processing techniques.

In the second step, duplicate EAC-CPF records that describe the same person, family, or corporate body are automatically matched and merged into a single record. This process presents a challenge for SNAC. It’s not unusual for people and corporate bodies to use different names and leave records behind under those various names. Additionally, the EAD records, while structured, often contain text data created manually by humans. Consequently, names may contain misspellings or the same name may be entered in a variety of ways, such as “Twain, Mark, 1835-1910,” “Twain, Mark,” “Mark Twain,” or “M. Twain.” For more information on this step and further technical details about the entire SNAC project, see Larson and Janakiraman [Larson and Janakiraman 2011].

The third step involves matching the set of EAC-CPF records against authority records from the Library of Congress Name Authority File (LCNAF) and the Getty Vocabulary Program Union List of Artists’ Names (ULAN). Doing so will enable the project to set the authoritative form of a name as the primary name in the EAC-CPF record as well as incorporating a list of alternative forms. Also, biographical/historical data found in the ULAN files will be added to the data in the EAC-CPF record to form a more complete description of the person, family, or corporate body.

As of August 2014, SNAC has derived more than 2.6 million EAC-CPF records by extracting data from 2.2 million WorldCat archival descriptions and nearly 300,000 British Library authority records. Future phases of development expect to derive data from nearly 190,000 more descriptions of historical collections from other government and academic archives and libraries [IATH 2014]. SNAC’s prototype access system and the radial graph visualization tool are the current examples, but more may be developed. Perhaps, in time, such tools will come to be considered as essential to humanities research infrastructure as the earlier layers of the stack that support them.

This case study has discussed the development of cyberinfrastructure at the level of building tools to enhance research. I used SNAC as an example because it allowed me to illustrate several variables that ought to be considered when designing a research tool for humanities cyberinfrastructure: size and scope of the project, members of the project’s team, and choice of technologies. SNAC aims to be a project of great size focused on a narrow mission — creating a network of archival data. To meet the specific research needs of its anticipated community of users SNAC will leverage the power of its data to provide access to research materials through integration with existing infrastructure. This goal could not be achieved without a diverse group of collaborators, from technology specialists to humanists, each representing the various domains of expertise that contribute to the project. This variety of influence on SNAC’s design contributes to decision-making that weighs the merits of traditional infrastructure and advanced cyberinfrastructure. SNAC adopts and develops technologies that best meet the needs of humanities researchers while pushing the limits of what is possible. One key way designers of SNAC ensure that the needs of current scholars are met is to heed the heritage of the project’s infrastructure stack. The creators of SNAC have understood the infrastructure of the past and present, identified what has served users well and which aspects need improvement, and thought creatively about how to augment existing services and solve current problems. Further examples of new cyberinfrastructure are likely to evolve naturally from the layers of existing infrastructure in this way as opportunities to build new layers are realized in the future, just as the infrastructure layers of the past were once new solutions to old research challenges. Research infrastructure both reflects and inspires humanities thinking and methods. As old obstacles to research crumble through the development of new infrastructure, new methods evolve and generate new challenges to bedevil researchers. The developers of infrastructure respond with technology and ingenuity, creating new services to meet these challenges. The cycle continues to this day, where digital technologies are inspiring new methods of research and new tools are being designed to facilitate, enhance, and accelerate those methods while eventually setting the stage for the next evolution in humanities thinking.

While I have discussed mainly tool building, cyberinfrastructure is much more than that. Cyberinfrastructure is the digital manifestation of a research culture. However, even while designing cyberinfrastructure at this grand scale, some of the variables that contribute to the success of a single tool may also contribute to successful shaping of a digital culture. SNAC can serve as an illustrative example in this regard, too. To better serve the entire research community SNAC intends to grow from the size of a research project to a self-sustaining national program. The Institute for Museum and Library Services (IMLS) has funded a project named Building a National Archival Authorities Infrastructure to help meet this goal. A proper mix of collaborators that represent the breadth of the project’s stakeholders is essential to bring this plan to fruition. The IMLS funding will support a meeting of leaders in the archive, library, museum, scholarly, and funding communities to determine the requirements of building a sustainable National Archival Authorities Cooperative (NAAC) [IATH n.d. d]. This cooperative would go far beyond SNAC in scale, but share the same narrow scope as SNAC. It intends to do a few things very well, but at a national scale: provide integrated access to the records held in government, business, and research archives across the United States as well as access to the social networks of the people whose lives are documented in those records. At this level of cyberinfrastructure building, researching new technology plays less of a role than developing new policy recommendations for the business, governing, and technical requirements for NAAC. However, a great deal of emphasis is placed on the top of SNAC’s infrastructure stack, the EAC-CPF standard [IATH n.d. e]. Writing about EAC-CPF several years ago, Pitti predicted that this “standard for creator description will facilitate building international, national, regional, and institutional biographical and historical databases that can serve as resources”  [Pitti 2004]. To that end, the IMLS funding will also support workshops and scholarships to attend them in which EAC-CPF will be taught to increase understanding and expertise in archival authority control. Broad adoption of this standard by the archival community is a necessary component to building the immense collection of data which tools like those developed in the SNAC project require in order to become a major contribution to cyberinfrastructure.

As progress continues with SNAC, NAAC, and other related research efforts, it appears that tools like those discussed in this case study will continue to grow and serve their intended community of scholars. In time, perhaps new projects will emerge to fill gaps in other disciplines. As this happens, we will get closer and closer to the goal set forth by the ACLS [ACLS 2006] report on cyberinfrastructure to create an integrated digital representation of the cultural record, connect its disparate parts and make the resulting whole more available to one and all, over the network.