Bhakti Virtual Archive

Development History


The idea for BHAVA began in 2014, shortly after the Regional Bhakti Scholars Network was founded. This new network inspired us to bring together technology and collective expertise to create a resource that would represent and serve the RBSN’s broad interests. At first, we (Jon Keune and Gil Ben-Herut) imagined this resource as a simple, large database of bibliographic references. We gave it a transparent name—the Connected Bhakti Bibliographies Database (CBBD). The CBBD was to consolidate the individual bibliographies that RBSN members had already collected in their own research and “connect” the diverse references by cataloging each item by using a specialized closed vocabulary of terms.

At the outset, we anticipated that people’s shared excitement and interest would suffice to carry the project forward. We thus sought out collaborators who could curate bibliographies that covered strategic range of languages, regions, and topics. A major goal throughout the project has always been to design the project in a way that drew on collaborators’ expertise and preexisting bibliographies, entailed some extra work in terms of cataloging materials the collaborators were interested in anyway, but would not become so burdensome and tedious that people would withdraw. We sought to discern everyone’s comfort in using either EndNote or Zotero to collect and catalog references, since many of us already used these applications. We had only begun to work out how to combine individuals’ bibliographies and into a consolidated, searchable resource. We anticipated doing this in our spare time, with no financial support. The CBBD was to be a shoestring effort by and for RBSN collaborators, using resources that were already at hand.

Digital Humanities and professional academic metrics

As we developed the project in 2014-2016, we both took up faculty positions at R1 universities and began reckoning with professional expectations of research and publication. Collaborators were also in positions and situations that prioritized commonly acknowledged research output such as monographs and journal articles. Our shoestring idea for a consolidated bibliography didn’t fit these metrics well and thus risked not “counting” toward professional advancement.

These professional concerns for ourselves and the collaborators, led us to scale up and formalize the project within the larger world of the Digital Humanities (DH). This was new territory for us, but we appreciated the scholarly rigor and thoroughness of the growing field. As we became acclimated to the institutional cultural of DH, we recognized how crucial questions about metadata and data management could bring our project it into alignment with established standards. But since DH was still a relatively new academic enterprise, we had to ensure that this project would be “legible” to review committees, without assuming that they valued DH and our project as we did. The way to manage this uncertainty, we were told, was to seek a major external grant.

Involvement with DH specialists at our respective universities sharpened the project immensely and introduced new possibilities for connecting to other projects and resources. Intellectually and substantively, this project matured a great deal as we talked with digital humanists, data librarians, and metadata specialists. Seeking a grant seemed like the logical next step, to validate our efforts externally and to incentivize us and our collaborators to complete the project. In conversation with research administrators, it seemed like two grant programs in the National Endowment for the Humanities were potential fits for this project.

Challenges and adaptations

Between 2017 and 2020, we adapted the project to fit the guidelines of the NEH Digital Advancement Grant and NEH Humanities Collections and Reference Resources grant programs. This involved enhancing the project in several ways—application development, cataloging standards and metadata, data management, and expanded target audience. The new vision for the project led to a new name: the Bhakti Virtual Archive or BHAVA, an acronym that corresponds to a key term within many bhakti traditions (bhāva means strong feeling, faith, and even bhakti itself).

In retrospect, although our vision for BHAVA grew and improved in the process, we spent a good deal more time in the past few years writing grant proposals than developing BHAVA itself. Two grants from DH@MSU and the American Library Association’s Carnegie-Whitney program supported us in developing components of BHAVA, but it would require a major grant to bring BHAVA to fruition. This major grant has eluded us so far. Reviewer feedback from multiple NEH proposals revealed that BHAVA is not as good a fit with the grant programs as we had hoped; reviewers seemed to prioritize technological innovation over intellectual contributions. The fact that BHAVA deals extensively with bibliography was also a drawback, especially since many references were to materials that existed physically in western libraries or virtually behind paywalls and thus could not be completely free to users. (And to produce such completely open knowledge would require a revolution in academic publishing or carrying out a much larger and more massively funded collaborative research project to create new knowledge.) After revising the proposal to address one year’s reviewer comments, we discovered a subsequent year’s committee criticizing exactly that change and suggesting the opposite of the first committee.

BHAVA has continued to develop through this all, but it has been disheartening to spend more time on grant writing than on carrying out the original project that inspired and excited us in the first place. So, in the summer of 2020, as we entered more professionally stable stages in our careers, we decided to stop applying to the NEH programs, free BHAVA from the confines of those grant program guidelines, and return to sleeker, more basic approach to accomplish BHAVA’s intellectual goals more directly. In doing so, the excitement and inspiration that we originally felt in 2014 has returned, and we look forward to this new, freer phase in development.

Technical decisions

We made some pivotal technical decisions about BHAVA’s development over the years. From the beginning, the Faceted Classification System was a central component of BHAVA. The major intellectual contribution of the project hinges on this System, which enmeshes every individual bibliographic reference into a single, searchable system. It is here that DH technology makes the most of the RBSN’s collective expertise, as bibliography curators use the System to catalog each record extensively. After a bibliographic record is indexed using the System classification fields, it becomes possible to find it when searching for this field. For example, a search of “Approaches=Hagiography” will return references that curators indexed as having this field value. Indexing each bibliographic reference with custom fields from the Faceted Classification System makes it possible to identify and locate patterns in bhakti scholarship across different periods, regions, languages, and traditions.

We initially thought about implementing this system using a shared EndNote or Zotero library. However, neither platform supports adding custom {field: value} data to a bibliography record, only string based notes. Storing indexed data in the Notes field can hinder possible searches, such as by slowing down the search and complicated the use of multiple values of the same field. These platforms also do not offer advanced search mechanisms or display options.

To address this problem, we considered aggregating individual curators’ bibliographies in EndNote and then exporting them to a custom bibliography management system that we would develop using Drupal. This approach would have allowed great flexibility in every aspect (database, search interface, etc.), but it had several major drawbacks: high development costs, dependence on the Drupal framework and its maintenance demands, and lack of standardized support for managing bibliographical records.

This led us to rethink the project’s architecture, resetting our priorities to a maximal utilization of existing technologies and platforms and a minimal reliance on costly and nonstandard development. We thus decided to use the widely supported Zotero Web Group library framework to manage our bibliographic database, and we settled on WordPress as BHAVA’s platform, because of its quick development time, lighter maintenance demands, and open architecture for adding code.

Embedding the Faceted Classification System into the bibliography library posed a major challenge since Zotero does not allow adding custom {field: value} dictionary/hash tables to bibliographic items. We explored several existing applications as workarounds, but we found them either too maintenance intensive or incapable of the kind of searching we needed to do:

  • The WordPress plugin ZotPress is an open source tool that offers an easy interface from a WordPress page to a Zotero library. Many sites use it, and it supported by a small community of developers. Despite its ease of use, the plugin does not have the option of adding custom search fields to the Zotero bibliographies and has limited search and display options.
  • Kerko is a robust framework for displaying and searching a Zotero Library with customizable search fields. It is based on a Python / Flask architecture and therefore requires system admin deployment and management.
  • Other projects (Wikindx, zsyncdown) offer customizable bibliography-managed websites, but their deployment also requires server admin. In addition, they do not support an external and scalable bibliographic engine such as that provided by Zotero.

Ultimately, we created our own PHP-based WordPress plugin, using funds from an ALA Carnegie-Whitney grant. The plugin uses the local MySQL database within WordPress to store tables of faceted terms, along with Zotero item IDs and their corresponding classification fields. Querying on faceted search terms within WordPress finds the relevant item IDs that can then be used to fetch the full bibliographic records from Zotero through an API request. This approach allows for very fast queries, since it is carried out within the local database. In addition, internal Linked Open Data attached to each of the faceted values will enable linking search results to external resources and repositories, when we reach that stage in project development.