Contextualized Integration of Digital Humanities Research: Using the NeMO Ontology of Digital Humanities Methods

The advent of digital infrastructures for arts and humanities research calls for deeper understanding of how humanists work with digital resources, tools and services as they engage with different aspects of research activity: from capturing, encoding, and publishing scholarly data to analyzing, visualizing, interpreting and communicating data and research argumentation to co-workers and readers. Digitally enabled scholarly work, and the integration of digital content, tools and methods, present not only commonalities but also differences across disciplines, methodological traditions, and communities of researchers. A significant challenge in providing integrated access to disparate digital humanities (DH) resources and, more broadly, in supporting digitally-enabled humanities research, lies in empirically capturing the context of use of digital content, methods and tools. This paper presents recent and ongoing work on the development of NeMO, an ontology of digital methods in the humanities, and its deployment for the development of a knowledge base on scholarly work.

Several attempts have been made to develop a conceptual framework for DH in practice. In 2008, a project funded by the UK’s Arts and Humanities Research Council, the AHRC ICT Methods Network, based at King’s College, London, developed a taxonomy of digital methods in the arts and humanities. This was the basis for the classification of over 200 digital humanities projects funded by the AHRC in the online resource arts-humanities.net. This taxonomy was subsequently modified by Oxford University as the basis for the classification of digital humanities initiatives at the University ( Digital Humanities at Oxford). Other initiatives to build a taxonomy of Digital Humanities include TaDiRAH and DH Commons. From 2011 to 2015 the European Science Foundation funded the Network for Digital Humanities in the Arts and Humanities (NeDiMAH). This Network was established to develop a better understanding of the practice of DH across Europe, and ran over 40 activities structured around key methodological areas in the humanities (digital representations of space and time; visualisation; linked data; creating and using large scale corpora; and creating editions). Through these activities, NeDiMAH gathered a snapshot of the practice of digital humanities in Europe, and the impact of digital methods on research. A key output of NeDiMAH is NeMO: the NeDiMAH Ontology of Digital Methods in the Arts and Humanities. This ontology of digital methods in the humanities has been built as a framework for understanding not just the use of digital methods, but also their relationship to digital content and tools. The development of an ontology, rather than a taxonomy, stands in recognition of the complexity of the digital humanities landscape, the interdisciplinarity of the field, and the dependencies that impact the use of digital methods in research.

NeMO was developed by the Digital Curation Unit (DCU), IMIS-Athena Research Centre, in collaboration with NeDiMAH, as a conceptual framework capable of representing scholarly work in the humanities, addressing aspects of intentionality and capturing the diverse associations between research actors and their goals, activities undertaken, methods employed, resources and tools used, and outputs produced, with the aim of obtaining semantically rich structured representations of scholarly work. It is grounded on earlier empirical research through semi-structured interviews with scholars from across Europe, which focused on analysing their research practices and capturing the resulting information requirements for research infrastructures. Its intellectual foundations lie in earlier work of the DCU on conceptualizing and modelling scholarly activity in the arts and humanities, conducted within the Preparing DARIAH, DYAS / DARIAH-GR, and EHRI projects, and manifested in the Scholarly Research Activity Model (SRAM), an ontological representation of scholarly information activity drawing from cultural-historical activity theory and process modelling, and compatible with CIDOC’s Conceptual Reference Model ( CIDOC CRM, ISO 21127:2006).

Architecturally, NeMO adopts a three layer structure, spanning from abstract/general to concrete/special concepts, to provide a flexible framework suitable to the multidisciplinarity of DH. Its top tier concepts (Actor, Activity, Object) provide a general reasoning frame, and function as semantic links to reference ontologies such as CIDOC CRM. These abstract notions are specialized in the second layer by way of domain-specific concepts covering every aspect of scholarly work: Methods employed in activities of various degrees of complexity or taught in Courses, Tools used, Information Resources taken as input or produced as output, Groups/Organizations or Persons participating in various roles, Goals addressed, Topics covered, etc. Furthermore, in this second layer, several semantic relations capturing the context of the aforementioned core concepts allow for modeling scholarly work through four complementary perspectives: (1) Process-related, centered around the concept of Activity and capturing temporal and spatial aspects; (2) Methodological, centered around the Method concept and capturing "how" aspects; (3) Agency-related, centered around the Actor and Goal concepts and capturing "who" and "why" aspects; and (4) Resource-related, centered around the Information Resource concept and covering "what" aspects of scholarly work.

Ιn the third layer of NeMO, fine-grained notions supporting domain-specific detailed descriptions are represented as specializations of second layer concepts. Respective vocabularies are organized as SKOS thesauri. More specifically, controlled vocabularies of lexical terms are structured hierarchically under the concepts of ActivityType, MediaType, InformationResourceType, TopicKeyword, ActorRole, SchoolOfThought and Discipline, which are specializations of the Type concept of the second layer, and are used for characterization/classification in parallel to ontological classification. The role of these taxonomies is, thus, twofold: (1) as a vocabulary of terms that can be used for flexible tagging of the objects of interest; (2) as entry points for the alignment, or mapping, of terms from NeMO to terms from other existing taxonomies. The latter enables integration with related work, as well as effective use of these taxonomies as documentation instruments or entry points for content in NeMO knowledge bases. For instance, the ActivityType taxonomy is organized in five hierarchies roughly corresponding to Unsworth's "cholarly primitives", and offers a flexible tagging system for modelling the intentionality of actors, scope adherence of activities, or purpose of use of tools and methods. On the other hand, mappings through broader/narrower term relations from the ActivityType terms to terms of other method taxonomies, including TaDiRAH, Oxford ICT and DH Commons, allow using those taxonomies transparently within NeMO.

The development of NeMO contributes to the work of the Digital Methods and Practices Observatory Working Group of DARIAH (DiMPO), as well as of Europeana Research within the Europeana Cloud project, providing an intellectual foundation for the analysis of evidence on arts and humanities scholarly activities and needs with regard to digital resource access across Europe. The relevance of the ontology to the DH community was validated through interviews and web surveys, to elicit information needs and patterns in working practices among humanities researchers, as well as two workshops in which these patterns were explored through use cases contributed by researchers. The evidence collected demonstrates that NeMO addresses adequately the knowledge representation needs manifested there. A variety of complex associative queries articulated by researchers in these workshops were also collected, demonstrating the potential of NeMO as an effective mechanism for information extraction and reasoning with regard to the use of digital resources in scholarly work; queries were encoded in SPARQL, a language appropriate for exploiting the serialization of NeMO in RDF Schema (RDFS), thus highlighting the benefits of its potential use as a knowledge base schema.

A prototype implementation of the above functionalities provides an easy to use demonstration of NeMO's potential. Users can articulate queries in structured English, without prior knowledge of any specific query language, using an intuitive user interface offering dynamic feedback of suggestions based on the conceptual schema. Input to the knowledge base is also supported by the same mechanism, guiding the user according to relations and classes provided by the model. A use case will be presented by way of example.

In sum, NeMO offers a well-founded conceptualization of scholarly work, which can function as schema for a knowledge base containing information on scholarly research activity, including goals, actors, methods, tools and resources involved. NeMO can thus be useful to researchers by (a) helping them find information on earlier work relevant for their own research; (b) supporting goal-oriented organization of research work; (c) facilitating the discovery of yet uncharted paths with regard to resources, tools and methods suitable for particular contexts; and, (d) promoting networking among researchers with common interests. Additional benefits for research groups include support for better project planning by explicitly representing links between goals, actors, activities, methods, resources and tools, as well as assistance for discovering methodological trends, future directions and promising research ideas. Furthermore, for funding organisations, research councils, etc., NeMO can (a) provide a bird’s eye view of funded scholarly activities; (b) enable the systematic documentation of research projects; (c) support evaluation of proposals, monitoring and control of project work, and validation of project outcomes.

Planned improvements include the development of mechanisms for providing recommendations based on semantically related instances and for the semi-automatic population of the knowledge base, as well as specialization of core classes and addition of new terms in Type taxonomies to reflect developments in DH scholarship.