Abstract
This paper provides both an update concerning the setting up of the European DARIAH infrastructure and a series of strong action lines related to the development of a data-centred strategy for the humanities in the coming years. In particular we tackle various aspects of data management: data hosting, the setting up of a DARIAH seal of approval, the establishment of a charter between cultural heritage institutions and scholars, and finally, a specific view on certification mechanisms for data.
Zusammenfassung
Der Beitrag gibt einen aktuellen Überblick über die Entwicklungen der europäischen DARIAH-Infrastruktur, vor allem hinsichtlich verschiedener Handlungsstränge, die strategisch auf Forschungsdatenzentren für die Geisteswissenschaften ausgerichtet sind. Insbesondere werden unterschiedliche Aspekte des Forschungsdatenmanagements betrachtet: Datenhosting, die Entwicklung eines DARIAH Seal of Approval, die Etablierung einer Charter zwischen Institutionen des kulturellen Erbes und Forschenden sowie weiterführend einen Zertifizierungsmechanismus für Daten.
1 Setting up a European Infrastructure – 10 years back
The history of DARIAH began in January 2006 when representatives from four European institutions[1] met to identify how they could join efforts in providing services to the research communities they served, with a strong focus on the humanities. The idea behind this meeting was to work towards a consortium of institutions, which would ensure long-term sustainability of the underlying infrastructure and a strong political voice vis-à-vis the EU.
DARIAH was thus put together as a top-down initiative of scientific information institutions, each having a duty to provide services to their respective research communities. It was conceived without the research communities themselves. But we would analyse this as a very beneficial factor since it gave us flexibility with regard to the actual scholarly coverage of our activities.
In the following years, the DARIAH model was completely inverted so that it has become a bottom-up organisation based upon active members and communities. In this context, there is a need at the highest level of the DARIAH management to reflect upon the complexity of the DARIAH landscape and identify the major priorities that are likely to be impacts on the scholarly communities in the future.
The present paper intends to be a contribution to this process by tackling a number of major issues related to the quality and status of digital information in scholarly work, and the possible role DARIAH could play in setting up and pursuing the corresponding agenda. After a quick presentation of DARIAH as it stands today together with some of the current achievements, we will focus on some concrete proposals that we would like to push forward in the coming period in order to achieve better services, availability, quality and scholarly recognition in relation to scholarly data sets. We would like to demonstrate that a series of parallel issues have to be dealt with in a coherent way if we want to be both successful in our endeavours and useful to researchers.
2 A quick glance at the current state of DARIAH ERIC
2.1 Missions and organisation
DARIAH has been set up at European level as an ERIC[2], the official European legal framework for research infrastructures, based upon a consortium of states that agree to support the infrastructure for a long-term period. In this context the main missions of DARIAH were conceived so that they would be an essential instrument to accompany the move towards digital methods in the arts and humanities. More precisely, DARIAH is responsible for achieving the following missions:
Identify infrastructural needs of scholarly communities and provide to help fulfil those needs;
Coordinate national contributions in establishing sustainable digital services for the arts and humanities;
Contribute to the establishment of national infrastructural roadmaps;
Participate in setting up the European agenda for infrastructures in Europe.
In order to achieve this, several organisational instruments have been set up in DARIAH:
The general assembly validates the budget and the general organisation of DARIAH;
The scientific advisory board accompanies DARIAH in setting up its scientific agenda;
The national coordinator committee, with active representatives from all countries, is the place where members compare and synchronise their national priorities;
The joint research committee coordinates and synthesizes the national contributions to offer concrete services for the community as a whole, in particular through its main instrument, the working groups, to which we will return below.
This complex environment, which is needed to ensure maximum communication and concertation at European level, is managed by the DARIAH coordination office.
2.2 A complex membership structure and a complex landscape
From the initial small group of countries represented in the first phase of DARIAH, the setting up of the ERIC consortium and its submission to the European Union have created an impetus for more countries to join DARIAH, and its following founder member countries: Austria, Belgium, Croatia, Cyprus, Denmark, France, Germany, Greece, Ireland, Italy, Luxembourg, Malta, The Netherlands, Poland, Serbia, and Slovenia.
In the course of the first year, two additional countries, namely Poland and Portugal, joined the membership, as well as five institutions form Switzerland as associated members. This last example reflects the fact that, depending on the national setting, actual participation in DARIAH can be a step-by-step process, as long as the various following essential components are in place:
A strong network of academic institutions that are willing to work together under a clear leadership;
Explicitly stated political support at Ministry level that ensures the (midterm) viability of the DARIAH participation from a national point of view.
As a whole, we can observe that the DARIAH Membership is based upon quite a variety of models across the different countries, which in turn depend on the following factors:
More or less strong national involvement in the development of digital services in the humanities depending on the actual existence of a national roadmap and thus capacity;
Various sustainability models based upon a project based funding scheme or the establishment of national service centres that act as counterparts to DARIAH at European level;
The balance between research and culture, depending on whether the preservation and dissemination of cultural heritage is seen as part of the DARIAH agenda or not;
Last but not least, the actual development of research funding programs to foster digitally based research projects.
This quick overview is in itself a good indicator of the challenges we face every day in fulfilling our mission, still we shall see in the following section that we have already managed to achieve a significant number of results.
2.3 First achievements
DARIAH is designed from the ground up to create and share resources for the digitally enabled arts and humanities, for the benefit of all. The chief of these resources are funded and sponsored on the national level, and submitted centrally as part of the formal membership contribution. Presently, there exists therefore a vast library of tools, services, software, project outcomes, workshop proceedings, as well as work on standards and teaching resources. One of the principal aims of DARIAH is to promote and facilitate the re-use of such materials, in order to accelerate new research, enable a more responsive and open publishing environment for humanities data, allow collaborations to be concluded more efficiently, with the overall goal of supporting not just digital services and complex digital objects in arts and humanities research but to broaden and deepen the adoption of digital methods and digital workflows amongst humanities scholars. DARIAH is currently working on refinements to its collation of these so-called “in-kind contributions” from its members, in order to ensure very precise interrogation through a single portal. DARIAH is also in the position of revising the criteria for evaluating this wealth of resources, which will be touched on in terms of the proposed DARIAH Seal of Approval, below.
Another major advancement in DARIAH’s operations recently, mentioned above as the ‘bottom-up’ approach, has been the establishment of working groups as the means, not only of offering defined fora for international collaboration but placing the research interests of the community that we serve at the heart of realising our strategic goals in DARIAH. This means that, in conjunction with our Scientific Board, DARIAH is extremely well plugged in to researcher communities, and hence the tools, services, and resources that it is broadly defining and helping to deliver for the future are tightly allied with cutting-edge thinking in this area. The working groups will also be instrumental in making the DARIAH Seal of Approval happen on a thematic level, with different aspects of the Seal (authentication standards, TEI protocols, etc.) being allocated to review and assent by the representative expertise of the appropriate working group. This will mean that those contributing to DARIAH will be able to concentrate their efforts in a particular area, be it authentication for example, or standards for annotation and publishing digital editions, knowing that the assessment will be undertaken by specialised academic peers within the overall progressive framework of DARIAH’s goals and activities.
Additionally, DARIAH, both as an integrating and membership body, has also either centrally collated, facilitated or helped create a range of services, for application across a broad spectrum of humanities research. These extend from examples such as OpenATLAS[3a], a database system for working with archaeological, historical and spatial data, to Ramses, an annotated corpus of Late Egyptian Texts, via Semantic Topological Notes (SemToNotes[3]), a topological image annotation and image retrieval system; GINCO,[4] a software platform for the management and distribution of scientific and technical terminologies, and SYNTHESIS,[5] an information system for the administration and promotion of cultural assets. In terms of this category of its “in-kinds” portfolio alone (there are 12 in all) the multifaceted DARIAH community has produced some 70 distinct platforms, services, tools, applications, and software environments to diversify and deepen the use of new digital techniques amongst humanities scholars.
As can be seen from the above services, there is necessarily a strong and intrinsic bond between the kinds of research arts and humanities communities undertake, and primary collections. In this respect, certain of the larger-scale services are also fully-fledged DARIAH affiliated projects. These projects, which have co-arisen with DARIAH and its impact over the past few years, but especially since 2012, have redeployed resources from the wider DARIAH setting and also the principles promoted by DARIAH centrally: Open Access, data-sharing, open standards, public humanities, and so forth. Having vision and a strong ethical commitment to open research in this area is as important to DARIAH as providing the practical and innovative means of doing digitally-enabled research. Thus EHRI as an affiliated project has worked in a novel way to bring the public actively into its development and the realisation of meaningful virtual relationships between highly dispersed and multilingual collections, and CENDARI[6] has not only repurposed DARIAH services, such as NERD for manuscript annotation, but has consistently worked towards researchers and collection holders making their work more open.
One of the domains where DARIAH has been particularly active since even before its official start is that of training scholars in digital methods. Under the impulse of the Virtual Competence Centre dedicated to research and education (VCC2), but also in conjunction with a number of DARIAH-affiliated projects, various activities have led to concrete support for humanities scholars, especially where they have initially to be self-taught in digital methods:
Numerous summer schools have been organized in the past five years targeting a broad range of scholarly communities (historians, archaeologists, philologists) or covering general aspects of the digital humanities methods from an editorial (publication) or technical (data representation, programming) point of view;
A reference Course Registry[7] has been set up and put into production, which gathers descriptions of the multiple curricula in European higher education institutions that are open to humanities students and cover the skills needed in and for the digital humanities domain;
Two flagship projects have received support at European level in this area. The first of these is DiXiT[8]. The Marie Curie Network has coordinated the most prominent centres in the domain of digital scholarly editions and has allowed numerous young scholars to benefit from a large spectrum of expertise from across Europe. More recently, the Erasmus+ program supported the #dariahTeach[9] project, which compiles and creates re-usable (and open access!) teaching materials for the digital arts and humanities.
The strong interest and support makes it clear that one of the many futures for DARIAH will intrinsically be to transmitting expertise within the digital arts and humanities.
With this general presentation of ongoing activities and results, we hope we have conveyed the general dynamics that exist within DARIAH. Still, there is a sense for most of us that DARIAH should have a vision of where it should prioritize its intended impacts on scholarly communities. The rest of this paper is intended to describe some components for such a vision, as we intend to implement it in the coming phase of DARIAH.
3 Ensuring data fluidity in the arts and humanities scholarly community
Even if it is an obvious point, DARIAH is, as a digital infrastructure, a data-centred one, and can thus be differentiated from more physical infrastructures such as GEANT, EGI or the forthcoming E-RIHS (for cultural heritage analysis equipment). As a consequence all DARIAH activities, from training to the deployment of technical services, are targeted at facilitating the availability and exchange of relevant data for research in the arts and humanities. We have already described a number of consequences of this data-centred strategy for DARIAH[10] and so the intention here is to suggest and outline areas in which DARIAH should be particularly active in the short- and midterm, in order to deliver concrete benefit for and at the service of scholarly communities.
3.1 Hosting as a priority
An essential pillar of the open access movement in the last decades has been the setting up of a network of publication repositories, which has allowed various bodies[11] to offer strong services for the dissemination of scholarly papers online, under rigorous technical and editorial conditions. In some communities such as Computer Science or Physics this has facilitated a shift from the individual dissemination of scholarly content though web pages (and before that, private snail mails to colleagues) to a more coordinated approach that, among other things, has secured the preservation and long-term access to the corresponding content. There could be many lessons to be learned from this long history of the publication repository landscape, but we can at least identify how much the specificity of the scholarly object “paper” has had an impact at various levels on the way such repositories have been conceived.
If we now come back to the issue of making the dissemination of scholarly data in the arts and humanities more fluid, we cannot but observe that the main request that scholars express when they produce research data in conjunction with their research projects is to be able to identify a clear setting where such data can be hosted in a trusted way.
In a way, scholars’ expectations are at odds with what the landscape looks like at present, seen from the point of view of infrastructures. There is indeed a wealth of possible repository solutions. First there are a few European countries that have set up generic solutions for hosting research data in the humanities. Repositories such as Nakala[12] in France or EASY[13] in the Netherlands already offer robust services and have integrated thousands of datasets since their respective launches. There are even completely generic solutions such as D4Science[14], used for instance within the PARTHENOS project, that offer to record any kind of scholarly output from publications or reports to datasets in any scientific domain. On the opposite side of the spectrum, we see that highly specialized hosting possibilities exist in relation to initiatives that intend to serve specific object types or scholarly communities. In this category, we can name services such as MediHal[15] for images (for all scientific fields). Infrastructures dedicated to the management and/or hosting of digital editions, such as TextGrid[16] (covering the whole editorial workflow) or Tapas[17] (for TEI documents), or generic infrastructures dedicated to linguistic content such as Ortolang,[18] but which could as well host complete documentary corpora.
One of the essential priorities for DARIAH will be to be able to find the optimal compromise between generic and specific deployments for data repositories, but also to be able to deal with the complexity of pulling together hosting services at European level. There is indeed a major challenge in making sure that hosting platforms do not reflect the fragmented national picture in Europe whereby countries will be reluctant to host (and thus to pay for) data that is not issued by scholars from their own research community. We thus need to find a way, and probably business models, so that data hosting becomes transparent for DARIAH users while making sure that there is a fair and balanced distribution of costs. As addressed already in Blanke (2016), linking crowds (scholars) and clouds (host) will be the key to the generalized dissemination of digital content in the humanities.
3.2 Towards a DARIAH seal of approval
The notion of a DARIAH Seal of Approval[19] could be seen as tending towards being essentially repositories-centric, given that the majority of specifications in this general modus often relate only to data and the management of data per se, as requirements to be met, so as to comply with any technical demands of repository services. We think we need to go beyond this necessarily limited operational view and offer instead a set of broader reference features that may apply directly to the data sets and other research outputs themselves, and that will simultaneously integrate required production and provenance values (related to whom has produced it, where and under which conditions it is hosted) but also fundamentally spell out how the data as such could be further and reliably re-used and redeployed in entirely new research contexts. Thus we aim to go from a repository-centric view of data to a scholarly-centric one.
The challenge for humanities research in general in this area can be straightforwardly expressed: how to find the sweet spot between taking advantage of generic and therefore highly reusable digital tools, software and components on the one hand, and maintaining a necessarily high degree of academic freedom to pursue novel and hitherto inconceivable research questions these digital affordances altogether allow, on the other. In short, how can one systematically foster innovation? The meme of the laboratory has been in vogue within especially library and archival circles in recent years to signal data-driven and open humanities research, and it remains a powerful and apposite analogy. A lab without fresh and novel ideas to be tested is sterile, yet to be operational, it also has to be rigorously maintained and work against known quantities and standards. It is in this spirit that the DARIAH Seal of Approval should be read: designing coherence for and at the service of experimentation.
The DARIAH Seal of approval would work in two main directions – to acknowledge that tools, services and software produced by the DARIAH community would meet criteria that allowed for their greatest potential reuse, on the one hand, and certify those primary collections at item level with which researchers wish to work as capable of maximum enrichment and subsequent access. There would be a number of discrete areas, modelled on the current in-kind contribution categories that DARIAH allocates to its member submissions to the central repository of shareable resources.
Digital objects are used in a wide variety of different contexts. Therefore quality here relates, not to absolute values but to use values. The Seal of Approval would not so much require of collection holders that certain minimum standards as such would need to be maintained as that information about the appositeness of any digital object be clear, precise and available to be exported or harvested. Since minimum standards at a technical level change over time, with the introduction of more powerful hardware and software, these would in any case have to be revised. But holding to some basic principles (describing a collection or items within any collection as being available via a well-documented API, what programming languages had been used for any tool or service under which digital objects were being made available) would greatly enable scholars to scope, find and identify resources that would work within the framework of their own particular projects and research questions with far greater surety. Such clear statements would also make it far easier for academics being asked to do so, to ‘give back’ their enrichments or annotations in a manner that could more readily be taken up and utilised by the CHI in question that provided the original digital resource.
These would not be classic data curation standards (which it goes without saying would also be applied), but beyond data conformance we need to focus now on data performance. The fact is that re-usability is sustainability. Services and collections will only last as long, not as there is a need for them but to the extent that their use in new modes of arts and humanities research is smooth and unproblematic – and for this the knowledge that academics need is not so much technical as logistical and legal: can it be downloaded? What are the licence terms? Is it available in raw text, json, XML, has it been transposed to RDF, etc.? These conditions need to be applied not just to a wholesale collection metadata description beyond the data in question but intrinsically as part of the data itself. These logistical considerations as well as the technical ones, for data performance and conformance, will also be an element of a further DARIAH-related initiative, PARTHENOS, which is designed to ensure the integration of outputs from both CLARIN-ERIC and DARIAH-ERIC, and that a workable, overarching framework is correspondingly put in place. This all being said, what are the kinds of actions that the DARIAH Seal of Approval will presuppose?
3.2.1 The Autonomous Data set
As stated in the above, data conformance has been tied very largely to the needs of data containers, not data users, i. e., data contexts. In that sense, we need to move towards describing data in ways that makes them independent of the service architecture or maintenance routines in which they find themselves. But what qualities would an autonomous data set have? How would it operate? There are some illustrative examples or prior suggestions of the qualities that autonomous data should have. In terms of curation, Manfred Thaller has already extensively described what he terms self-preserving objects, data that has “information (added) to the object, as is required to make it fit for processing on radically changed platforms within radically changed environments”.[20] Another example of data descriptors, and hence potentially greater meaningful access and reuse of digital resources, that are based not on procedural or technical parameters but on the intrinsic qualitative context of the data and objects involved, is CAMELOT, developed at the Oxford Digital Library. The categories of meaning in this data model would allow relationships between objects and their future re-use environments to be ascertained. So, not only are identifiers such as certain place types (academy, library) and person available within the model but aspects such as “Research funding, administration and projects, academic institution structure, scholarly activities, research communities, creative works, manifestations, instances, collections and aggregations...[and]...annotations”. These are the very types of scholarly-centric notions that should be applied to data as much as repository centric ones and we would want to encourage and embed as part of the scheme.
3.2.2 Authorship, provenance, reuse and citability
At the level of the monograph or article, authorship is a relatively unambiguous matter; despite some challenges, even the use and impact of articles with multiple authors can be overcome, in spite of such considerations as researchers moving between institutions due to career decisions, and research assessment value can still be properly allocated personally and institutionally. However, in the scenarios we are highlighting, with CHIs providing primary resources that researchers may build on or add to qualitatively, the research outputs may consist of annotations, other object enrichments, or algorithmic and code improvements. Additionally, if we imagine a much greater reciprocal arrangement between researcher and CHI, whereby the CHI gains greater recognition for the material that has been (often painstakingly) made available, the argument could be presented that in this context provenance equals authorship. Also, there remain issues around consistently and systematically being able to trace the trajectory of general archival use and access against the precise research impact of particular resources, especially where archives or special collections are offering new formats, such as video.[21] The opposite situation would have a positive sustainability effect on CHIs. The more it would be possible to demonstrate a clear relationship between use of digital (and analogue) archival resources, their citation and their formal research assessment and altmetric impacts, the easier it would be for CHIs to make the argument for funding based on evidence. This is crucial in a climate and at a juncture in digitally-enabled arts and humanities research, where to an extent CHIs would ideally want to take in researcher enrichments of primary digital resources, and researchers conversely are looking for ways of ensuring the long-term maintenance of their digital enhancements.
Overall then, we would require two basic things: on the one hand, much greater formal acknowledgement of and a way of auditing and assessing research outputs beyond the format of the book and article, so as to encourage amongst especially early career researchers further and more varied interactions with digital primary resources. On the other hand, a means of recognizing the ‘authorial’ hand of CHIs in facilitating access to and, in eventually gathering researchers’ digital enrichments, the conceptual and material enlargement of collections for further (re-)use.
There is also one more potential category of citability, and this refers to one particular, by now, highly dynamic aspect of digital object enrichment – user interactions, which, in their available, amalgamated form, are also a rich source of material in their own right for arts and humanities scholars. How would one cite a particular interrogation of such user-interaction data in a persistent way, such that it could be invoked reliably, thus bringing arts and humanities research closer to the overall research quality not simply of quantification but of reproducibility? As well as conformance to extant international formats for data referencing that should also become a significant aspect of the DARIAH Seal of Approval, there is also the potential for DARIAH to develop and set some new ones for citability.
3.2.3 Access
According to the above, therefore, conditions of access to the documents and collections would necessarily take into account the huge variety of data and metadata available through CHIs involved in these research processes and actively contributing therefore to research outputs. The CHI part of the Seal of Approval would mark out therefore significant ground for participation in the scheme. Regarding metadata, free (as in speech and beer) access to document metadata would be made available (including document enhancements). In terms of images, access to scans of any document or image according to current quality standards would be granted upon request. User interaction data would also be made available on request. The chief aspect governing this part of the Seal of Approval would therefore be licensing.
3.2.4 Licensing
In order to be eligible for any level of the DARIAH Seal of Approval, licence statements would have to be specified at all times and explicitly for each unit of the available content in question. The terms under which objects and metadata would be licensed to academics belonging to institutions that themselves were organisational signatories to the DARIAH Seal of Approval would need to be defined precisely from the outset and designed so as not to be susceptible to change. For this reason, it is recommended to implement from the beginning licences that will be as open as possible, which would prevent downstream use and re-use being more restrictive at a later point.
Default licences for CHI content, with further options depending on level of openness versus level of ‘recognition reward’ within the DARIAH Seal of Approval’ scheme desired (Bronze, Silver, Gold) would be described and recommended; similarly there would be default licences for enrichments. A neutral scale might be, rather than stipulate only certain instances of known concrete licences (such as Creative Commons) the categories of, for example, ‘not-stated’, ‘proprietary’, ‘restricted’, ‘open’. Concrete and current licence options could then be mapped onto these categories. However, the standard licence to be used would be an open one, with exceptions having to be actively applied for (and which naturally can be entirely legitimate, in the case of personal data and constraints placed on the use of material by depositors).
3.2.5 Technical setting
Finally, the guidelines shall contain some basic technical principles, which may apply either to the repository at large, a particular collection, or at the item level. As in the case of the Data Seal of Approval, features or exact technical requirements or expectations will not be made.[22] Instead, statements such as “Is the data represented and stored in formats that are compliant with international standards” or “Is the data optimally represented for long-term legibility of the content”, or “Does your digital content have PIDs” would be the kinds of guiding questions the DARIAH Seal of Approval would prompt in aspirants, to lead them through a series of realisable expectations that would allow institutions to apply for a level of assurance in the re-use of their digital collections that would be both achievable but also according to broadly acceptable and established practices. Again, it is possible to imagine that beyond the Seal of Approval guidelines, a decision tree tool could be made available that would enable institutions to correctly establish the level of re-use impact for their collections they would want to achieve, the requisite types of technical, logistical or licensing features their collections should intrinsically possess, and therefore what they would need to do in order to attain one or other of the levels of scholarly data assuredness within the DARIAH Seal of Approval.
Conversely, research institutions would be asked about what level of academic commitment they would want to achieve. However, as in the case of freely available metadata as a basic requirement of CHIs, there would be fundamentals to be met by research organisations wishing to comply with, and therefore obtain recognition from, the scheme. One of these would be an unerring commitment to offer any enrichment also back to the CHI that had provided the original digital resource on which they are based. The CHI institution would not be obliged to accept but an offer should be made, which could still be taken up in a sense, even where the local technical capacity might not exist immediately, through the provision of pointers to a reference host elsewhere. Another stipulation for acquiring even the basic level of scholarly data assuredness on the part of researchers/research organisations and CHIs would be the consistently applied use of extant authority files, whether these are for places, persons, companies, and so forth.
4 Data re-use charter
As important as it is, stating clear requirements regarding digital data and being able to offer solid hosting services to researchers still remains only one component in a wider landscape of data exchange mechanisms. The major issues at stake have to be addressed not as isolated aspects, but as a whole. With this objective in mind, we introduce in the following section a course of action aimed at moving ahead swiftly at the service of scholars, but also of cultural heritage and other institutions that act as service providers to the whole community.
Taking a look at the fundamental aspect of arts and humanities research, we see that it is mainly based on the analysis of what could be called more generally ‘human traces’. This concept covers a variety of possible concretions: artefacts, works of art, written documents of all sorts, recordings, etc. All of them have a historical dimension in common: they are inscribed in a (transmission, preservation) tradition. They also have in common that they are generally hosted in cultural heritage institutions, which themselves can be of various types and importance, ranging from national libraries and archives down to small, local museums endorsed by regional or city administrations. These institutions are not all bound to an identical organisational or institutional setting; on the contrary, they can fall within various domains (public, private, foundations, etc.).
Scholars exploring collections or exploiting single documents from CHIs all face similar and recurring problems when it comes to what they can do with the material they identify as relevant for their research. There is no generally valid rule as to how much they can quote, duplicate and furthermore republish in their scholarly work. This question extends to the various forms of research dissemination bound to arise from this work and which range from traditional publications to complex productions yielded from the institution’s content: catalogues, archival research guides, collections of images, transcriptions. From a cultural heritage point of view, it is not even clear what status such productions would have, even if, for instance, some institutions were enabled and encouraged to host the researchers’ by-products insofar as they may be seen as complementary resources to the original material.
The lack of a clear and comprehensive framework that could serve as a general baseline for interactions between scholars and cultural heritage institutions is a hindrance to both the development of further research projects based on highly valuable documentary collections, and the visibility of the institutions themselves as key actors in the research ecosystem.
The present initiative, launched by DARIAH-EU, but which aims at being widely inclusive of all organisations related to cultural heritage institutions, will provide such a framework. It is conceived as a win-win setting that has the strategic potential to act as a reference for all interactions between scholars and CHIs. It will be capable of relieving individual scholars or collection curators from the challenge of having to rule again and again on how to use and re-use documentary material on a case-by-case basis.
4.1 Relevant use cases: challenges and options
In the course of finalising the charter’s scope and phrasing, we need to gather relevant use cases that are bound to help us make sure to cover the majority of the scholarly community’s needs. We will only present here two standard cases that reflect the kind of hurdles commonly experienced. They will be enriched with further analyses, solutions and comparison with additional use cases in the near future.
4.2 Re-using iconographical material in publications
A typical example has occurred while finalizing the publication of the archival research guides[23] produced in the context of the Cendari project at the beginning of 2016. These guides have been mainly written by scholars from the medieval and First World War periods in the context of the transnational access program which allowed young scholars to dive into one specific theme from the corresponding period and gather historical evidence on the basis of the archival data made available to them through the Cendari architecture.
Although instruction had been given to them to check and quote the appropriate source for any iconographical material they would use in their research guides, we came across the situation where the actual status of the illustrative images was just unclear. In certain urgent cases, some further checks had to be carried out and consequently sometimes had to be taken out from the publication. In a way, the Cendari project was left in a situation where no clear intellectual copyright clearance could be activated for re-using even a single image taken, for instance, from the website of a cited archive.
In this case, we have to deal with a paradox: the commercial publication domain for instance has already set up a scheme that accounts for the “re-use of limited amounts of material from published works”.[24] The STM permission guidelines[25] state for instance the number of figures and tables that can be re-used in a publication without requiring any specific permission. It is noticeable too that some publishers also require to be notified of such use in any case.
So, why should scholars and infrastructures spend so much time checking such simple rights if eventually that material is not actually re-used because doubts nonetheless remain concerning the corresponding rights?
4.3 Digital edition from a physical textual source in a Cultural Heritage Institution (CHI)
For many scholars studying primary textual sources, the production of an edition is an intrinsic part of the research process. Digital editions reflect and accompany the evolution of this dimension of the research process.
Over the past years, the standards for digital scholarly editions have evolved towards a greater inclusion of the documentary basis within the editorial work itself. Many digital editions now comprise the transcription and (fine) annotation of a corpus of documents gathered from one or several libraries or archives – which are likely to become part of the edition itself. At each stage of his activity uncovering, unlocking, editing and exploiting these sources, the scholar is likely to face the following challenges and questions:
Looking at potential sources, and beyond the kind of searches one could make on a portal such as Europeana, he would need to know whether, for the CHI he is interested in, there exists a digital catalogue of the collections and items, and whether he is allowed to take partake of the totality of any related information for future use or publication;
Once he has identified required items in any library or archive, how can he be aware of the existence of digital surrogates or whether he is allowed to make and publish photographs or scans;
To what extent is the scan quality made available by the CHIs compatible with scholarly standards? When scan quality has improved noticeably, how can a researcher update his edition with regard to the scan quality (which requires the CHI to realize new scans and let the researcher use them according to the same terms as before), how can he negotiate the long-term cooperation with the CHI in this perspective?
How to merge archival, library and research metadata in a common dataset that makes the contribution of each partner visible and is quotable (question of multi-institutional enrichment, of hosting, of authority, of format compatibility and of the value of rich metadata in term of scholarly recognition);
Last but not least, how much can the source material be actually re-used within an online publication or even a printed object, as is still the case for some scholarly editions.
As we can see, and our list is probably not exhaustive, the level of potential complexity we reach here goes way beyond what an isolated scholar should need to understand and deal with. There is thus an urgency to provide a clearer setting for such scenarios.
4.4 Which actors for which partnership?
To pave the way towards the design of a common and generic data re-use charter for cultural heritage content, we first need to see what this would mean for each major category of stakeholder and in what form they could engage in such an endeavour, taking into account costs and benefits. The orientations sketched below are designed for the three main categories of stakeholders, namely cultural heritage institutions, scholars and hosting services. This analysis and the action lines drafted here are conceived as the basis for a wider dialogue that could and should enrich, and add greater precision to, what we suggest here.
4.4.1 Cultural heritage institutions
Cultural Heritage Institutions can be very diverse. History, mission or focus varies from one institution to the other, often involving strong regional specificities and legacy cultures, not to mention the impact of history. The charter aims at considering them in the generality of their function as curators of collections and objects in their physical form and as potential primary initiators of corresponding digital surrogates, from basic descriptions (catalogues of collections, metadata for specific objects) to more elaborate outputs (scans, 3D models, physical analyses, etc.).
In this context, we consider that cultural heritage institutions could engage in a dedicated course of action along the following lines:
Data delivery and services: Each institution should be in a position to describe by which means and in which formats it can provide access to its digital resources. Compliance with international standards and good practices such as the DARIAH Seal of Approval could be a major asset in this respect;
Access, use and re-use: The charter should offer the possibility for each institution to state its policy in this respect, and in particular to specify the access constraints (categories of users, fees) and the conditions (e. g. licences) under which a scholar can further publish any kind of content based upon material (documents, metadata) that it has initially made available;
Further hosting services: The institution should be encouraged to notify when it can provide specific data hosting possibilities for amended or enriched content. By agreeing in the charter principles as a data-hosting infrastructure, it can contribute further to the improvement of data curation. In such a setting, even more digital surrogates can be curated in a coherent environment, with the notable advantage that this includes the connection to the corresponding primary data.
The benefits for the undersigning institutions are numerous. The institutions would potentially gain higher visibility for their assets and be able to boast their support to researchers, with the potential impact on the institutional support they would thus get. They would also have a direct feedback loop from the researchers themselves. The advantages of such a direct communication with the scholarly community include a better understanding of possible digitisation strategies in relation to the further societal interest in specific collections or to emerging research areas. Finally, they would be linked to potential digital hosting institutions (in the case that they are themselves unable to take on this role, for infrastructural reasons or otherwise) with whom they could then directly set up coherent collaboration schemes.
4.4.2 Researchers
This charter aims at creating a direct relationship between scholars, CHIs and hosting infrastructures. This is why we contemplate that scholars could or even should sign in person independently from the institution for which they are working at the time they sign the charter. However, we would welcome academic institutions (departments, universities, research institution or funding agencies) wishing to sign the charter and even make it a requirement for their members or the projects they fund or host.
As we cannot expect that scholars will themselves have the technical or human capacities to implement complex technical parameters and because we want to engage as large a group of people and institutions as possible, we would limit the required commitment to the following aspects:
Compliance: Scholars must in all cases certify that they will comply with the use and re-use conditions stated by the CHI signatories of the charter, in order to create a solid relationship of pledged trust between them. This concerns in particular citation and licencing conditions, that have to be respected without exception;
Contribution to the open dissemination process: Material resulting from the scholar’s work shall be further distributed in the most sophisticated form possible, for instance as source XML-TEI files, and under such conditions that further re-use is straightforward, in particular for other signatories of the charter;
Priority of deposit to the source: When the cultural heritage institution from which the primary data is issued is offering this option, scholars shall offer to deposit their own enrichments there first;
Priority to hosting institution which has signed the charter: scholars should do their best to deposit their digital productions with institutions that are signatories to the charter.
The charter could probably cover, maybe as an option, a larger scope of objects. Typically, the scope could also comprise authorities such as gazetteers, prosopographies or bibliographies compiled from the primary sources used by the researchers. On this point DARIAH would want to consult with all relevant parties – researchers, CHIs and hosting institutions – when finalizing the charter.
The benefits of adhering to the data re-use charter embrace different dimensions for the researcher. They include easier access to cultural heritage collections, fewer legal uncertainties when looking into the use and dissemination of digital material, and a far greater level of trustworthiness in the preservation of his production. The researcher hence has a maximal interest in being part of this endeavour. We should especially consider advocating the charter to researchers in such a way that they will contribute, through their own expressed preferences, to work with and alongside both CHIs and repositories, in order to highlight the direct benefits also to the latter of adherence to the charter.
4.4.3 Data hosting infrastructures
Primary data can be hosted by CHIs or by Higher Education Institutions (HEIs) like universities, but they are in many cases curated by dedicated data hosting facilities as we have seen earlier in this paper. These institutions are equal partners in the Charter, alongside CHIs and scholars. They play a key role in guaranteeing the stability, the visibility and the long-time availability of the primary data. The engagement we would expect for hosting facilities is clearly more technical and should ensure a concrete implementation of the CHI-researcher relationship. We can think for instance of:
Proper definition of the scope of the hosting facility in terms of types of data, accepted or required formats, additional descriptors (meta-data) attached to the data sets;
Additional core services which are offered, in particular from the point of view of sustaining the content: long-term archiving, persistent identifiers, proper helpdesk, etc. Services (cf. SITAR, sustainability);
Access conditions to the archive: population being served (researchers, wider public), identification mechanisms, cost model (institution or project based for instance).
Again, we can identify numerous advantages for the corresponding facility: better mapping of the service landscape, exchange of expertise, collection completion, critical mass of content (behind the scenes, easier identification by governmental agencies of hosting sectors that require long-term support/sustainability).
4.5 Stakeholder inclusion
Devising and describing the content of such a charter will, however, be only a forerunner to ensuring that all meaningful stakeholders are identified and are enabled to actively participate in the process. Beyond DARIAH members themselves, with the prospect of gaining national recognition for this charter, we do need to include various transnational initiatives in the library (LIBER, FILA), archival (ICA) or scholarly domains (ADHO, EADH). Major aggregations such as Europeana would also need to be engaged at an early stage along with specific flagship collecting institutions that could be potential early signatories. We should also make sure that smaller institutions such as regional historical archives for instance are included in the initial discussions so that they can also express their specific constraints, which may be quite different from more national institutions.
Whatever the challenges, we can anticipate the not insignificant extent to which the workload currently shouldered by researchers, CHIs and repositories would be reduced if this data re-use charter were put in place. While the implementation effort is not expected to be minimal, the positive repercussions are projected to be everything but marginal.
5 Certification platforms
Putting what was described so far in perspective, let us turn again more specifically to the people that we are actually here to serve as infrastructure providers, namely the researchers themselves. This turn of perspective is required for at least one major reason: because what we present as priority actions for DARIAH may be conversely considered by researchers as further constraints on their work. These constraints would affect the way they should describe their data, but also where they could deposit and host them, and, finally, also the conditions determining how other actors could re-use the content they have produced. There are many possible arguments that could demonstrate the added value of a more rigorous and channelled setting for data management, in particular in providing a wealth of data that would facilitate the life of humanities scholars at large and allow them to make many new discoveries, hypotheses and comparisons if the data is just there, and at hand.
Still, in the short term, we need to set up the means to provide quick recognition for those who are spending a significant amount of scholarly time and effort on designing and distributing high quality research data sets. This means that we need certification mechanisms which scholars could boast in relation to their data sets: an evaluation setting that should not be based solely on technical evaluations, but should also reflect possible appreciation by colleagues, departments, home institutions, research councils and funders, and not least, any formal national research assessment frameworks. There is no reason why this should not work in the same way and to the same extent as we have experienced in the context of similar regimes of research quality review, impact and assessment in the realm of the scholarly paper.
The underlying concept is not completely new. It has emerged and crystallized in recent years around the notion of the data journal, instances of which have already appeared in various communities with, for instance, Geoscience Data Journal[26] or Earth System Science Data (ESSO)[27] in geosciences, the Biodiversity Data Journal[28] in biology or generic data journals such as CODAS A’s Data Science Journal.[29]
The actual impetus towards providing appropriate environments here has attracted major publishing companies to launch similar initiatives. From high-profile publishers such as Nature Publishing Group with Scientific Data to opportunistic ones such as Hindi publishing with its Datasets portfolio,[30] not forgetting the profit-making “scholarly” association ACS with its Journal of Chemical and Engineering Data,[31] the idea has already passed the stage of superficial interest, as it was still the case in the early 2000s or even with forerunners such as the Journal of Astronomical Data (JAD),[32] which pioneered this approach in 1995.
Despite these examples from other disciplines, there still does not seem to be a similar momentum in the humanities. This means that we dispose – precisely now – of a unique opportunity to scope what humanities scholars would need and therefore, how DARIAH as an infrastructure would help establish or facilitate data journals in the humanities and contribute to their overall quality. This would not only encompass technical and logistical aspects, but also a critical assessment of the notion of “journal” itself; that is, what essentially constitutes the necessary form of publication for arts and humanities scholars who are increasingly sophisticated in their digital approaches to research. What is the content of a ‘journal’ in this sense and therefore, what apparatus is needed for, say, the tracking of citations relating to a much wider scope of material than bibliographic – programmes, software, algorithms, methodologies and data itself – how such ways of describing the impact also of this type of output integrate seamlessly with academic career structures?
First, the selection process in scholarly journals that has developed over the years is mainly based upon a post-peer review publication process that has until now prevented many potentially interesting results and studies to be published in the first place.[33] Whereas this notion of selection originated in the, now digitally overcome, lack of available space in printed journals, this publication structure remains as a cultural relic in the scholarly ecoystem. If we want to ensure that all data sets in the humanities are actually made available, we should not aim at recreating print-world benchmarks and should think instead of crucially decoupling the actual publication (in the sense of “making public”) from the assessment stage, by going towards post-publication peer review.[34]
Going even further, we should not just consider peer-review as an acceptation/rejection-based mechanism, but more as a ranking or certification service that may rate a data set along various editorial, technical and scholarly dimensions that may reflect how complex a data set can be as such, as well as how multifarious its post-publication environments of re-use might be. Indeed, if we consider for instance the publication of a corpus of transcribed videos corresponding to the recording of human interactions in context, where various phenomena (prosodic, referential, etc.) have been marked up, we want to offer the possibility to assess different aspects such as the conditions under which the experiment has been set up, the technical quality of the recordings, the appropriateness and precision of the transcription and annotations as well as the compliance of the corpus to existing international standards.[35] Such a certification process may also be carried out in other conditions than the traditional, not to say baroque, blind peer-review principles. Reviews may be made open to readers,[36] as it is being implemented in the DH Commons Journal for instance. Reviewers may be selected by the submitter himself, since he is probably the most likely person to know those who are appropriately most knowledgeable in his field. Finally, the data set can be left open for comment by the community, under clear authentication conditions to prevent spamming, and so forth.
The various editorial issues that we have outlined above demonstrate the actual need to experiment and to offer a flexible platform for future communities to develop certification and research assessment environments. Above all, we think that the wide availability of data sets and associated material should not rely on specific business models such as the one we have observed in APC-based “open access” journals. Such models, for instance, make the choice of the actual licence attached to the content depend upon the fee structures that the author is required to navigate and satisfy. Openness should not be hostage to fortune and we believe that we need, as a complement to having a network of publicly-owned data repositories in the humanities, (future) data journal platforms that should also remain in the public sphere.[37]
This, however, is unlikely to be realised unless we can think of low-cost settings, which, in particular, do not duplicate existing data repository infrastructures. A possible answer is the establishment of overlay certification platforms which allow the data to be (first!) deposited in a given trusted data repository and whose reference will then be forwarded to the environment which adds a layer of peer- or community-review process, either of which may lead to a public assessment of the resource. This has been experimented with in the context of the Episciences platform.[38] This setting could serve as a freely available platform to set up experimental data journals. The steps towards establishing the necessary environments for such experiments can be itemized as follows:
Identify a domain in the humanities where data production represents an important aspect of scholarly work;
Identify criteria to assess the quality of such data production;
Select a core of data repositories where such data are or could be hosted;
Link these repositories to the Episciences platform through its simple OAI/PMH interface;
Motivate an enthusiastic editorial committee;
Develop a satisfying workflow.
Not all aspects are equally simple to implement, but an infrastructure like DARIAH has to take on its role by helping to stabilize and promote as many of the necessary underlying components as possible.
6 Ambitions for digital scholarship
As a whole, DARIAH is not just an administrative organisation coupled with a few European projects feeding its work plan. Since its very inception, it has reflected upon the tremendous digital revolution that is occurring in the humanities and that has to be accompanied by a strong coordination of all actors, and it has developed its roadmap on this basis. Such a vision requires the continuous identification of the missing elements in the landscape, of the stumbling blocks that are still barriers to the humanities being completely enabled to embrace the digital turn.
In the context of this article, we have highlighted a number of domains where we think that it is possible, thanks to the presence and capabilities of an infrastructure such as DARIAH on the side of humanities scholars, to initiate a deep change in the practices in digital scholarship. From very short-term aspects such as the repository landscape, to more sociologically complex proposals such as data journals, we think that it is necessary to propose and push a disruptive agenda, if only to make scholars reflect upon their own digital future.
References
Armbruster, Chris; Romary, Laurent (2010): Comparing Repository Types – Challenges and Barriers for Subject-Based Repositories, Research Repositories, National Repository Systems and Institutional Repositories in Serving Scholarly Communication. In: International Journal of Digital Library Systems. Verfügbar unter https://hal.inria.fr/inria-00465197.10.4018/jdls.2010100104Search in Google Scholar
Berthaud, Christine; Capelli, Laurent; Gustedt, Jens; Kirchner, Claude; Loiseau, Kevin (2014): EPISCIENCES – an overlay publication platform. Dr. Panayiota Polydoratou. In: ELPUB2014 – International Conference on Electronic Publishing, Jun 2014, Thessalonique, Greece. IOS Press, 78–87, 2014. Verfügbar unter http://www.ebooks.iospress.nl/publication/36552. Doi: 10.3233/978-1-61499-409-1-78. hal-01002815v2.Search in Google Scholar
Blanke, Tobias; Kristel, Conny; Romary, Laurent (2016): Crowds for Clouds: Recent Trends in Humanities Research Infrastructures. In: Benardou, Agiati; Champion, Erik; Dallas, Costis; Hughes, Lorna: Cultural Heritage Digital Tools and Infrastructures. Verfügbar unter 978-1-4724-4712-8. hal-01248562.Search in Google Scholar
Jones, Nicola (2013): Half of US clinical trials go unpublished. In: Nature, Verfügbar unter doi: 10.1038/nature.2013.14286.10.1038/nature.2013.14286Search in Google Scholar
Pöschl, Ulrich (2012): Multi-stage open peer review: scientific evaluation integrating the strengths of traditional peer review with the virtues of transparency and self-regulation. Front. Comput. Neurosci. Verfügbar unter 6:33. doi: 10.3389/fncom.2012.00033.10.3389/fncom.2012.00033Search in Google Scholar
Roberts, Peter (1999): Scholarly publishing, peer review and the Internet. In: First Monday, 4 (4): Verfügbar unter http://ojphi.org/ojs/index.php/fm/article/view/661/576.10.5210/fm.v4i4.661Search in Google Scholar
Romary, Laurent (2011): Partnerships, relationships and associated initiatives – Towards a strategic plan for DARIAH. DARIAH Research Report, R EU 4.3.1. DARIAH-DE. Verfügbar unter hal-01150112.Search in Google Scholar
Romary, Laurent (2015): Standards for language resources in ISO – Looking back at 13 fruitful years. In: edition – die Terminologiefachzeitschrift. Deutscher Terminologie-Tag e. V. (DTT) 2015. Verfügbar unter hal-01220925.Search in Google Scholar
Romary, Laurent; Armbruster, Chris (2010): Beyond Institutional Repositories. IN: International Journal of Digital Library Systems, January 2010. Verfügbar unter https://hal.archives-ouvertes.fr/hal-00399881.10.4018/jdls.2010102703Search in Google Scholar
Sinn, Donghee (2012): Impact of digital archival collections on historical research. In: Journal American Society of Information Sciences, 63, 1521–1537. Verfügbar unter doi: 10.1002/asi.22650.10.1002/asi.22650Search in Google Scholar
STM (2014): The STM permissions guidelines. STM association. http://www.stm-assoc.org/2015_12_10_STM_Permissions_Guidelines_2014.pdf.Search in Google Scholar
© 2016 Walter de Gruyter GmbH, Berlin/Boston
