|
Organizational structure
|
Size (Pool of DP employees measured in full time equivalents (FTE)) |
On average, EPs have the largest pool of employees, followed by GPs and BPs. By offering tools, EPs depend on personnel intensive work, like creating tools and maintaining corresponding data quality. GPs cover a wide scope of scientific domains resulting in a diverse user community, as well as high data volumes and use cases. BPs cover certain use cases in a specific domain with comparably low FTE values. |
|
Hierarchy (Arrangement of employees: High vs. low direct vertical coordination) |
In general, the higher the number of FTEs of a DP, the more likely it is that two or more teams exist, requiring more extensive vertical and horizontal coordination in the organization. While GPs often consist of more than two teams, BPs usually consist of one team only. |
|
Team diversity (Disciplinary background) |
All DPs rely on personnel with IT and computer science backgrounds to set up, maintain and advance the platform. The educational background of DPs personnel with high domain specificity (BP, EP) is often related to the domain-specific background. GPs favor a diverse educational background, as there is no anchoring within one particular domain. Employees with a diverse background signal a broader knowledge frontier, enabling the recombination of cross-domain knowledge, which is vital for innovative governance models. |
|
Boards and committees (Existence of advisory board, steering committee) |
The existence of advisory boards and steering committees is comparable between all DP types, and smaller BPs also have such entities. |
|
Technological infrastructure
|
State-of-the-art (Response to challenges related to technological change) |
EPs face challenges regarding the ambidexterity of simultaneously increasing data complexity and offering tools to users. Adapting existing tools to evolving data sets remains especially challenging. |
| For BPs, the degree of transparency remains important, as they facilitate data exchanges in specific data domains by functioning as a major storage space. Current challenges experienced by BPs comprise standardizing and expanding (meta) data, as well as data duplication. |
| The technologies of GPs are designed for metadata and data of many or even all domains, so that the platform can be linked to other (external) organizations. Thus, the architecture must be designed to link the scattered components of the infrastructure (e.g., from researchers, libraries, and journals) making APIs the key gateway technology for GPs. Challenges include storage space, covering individual use cases, and the identification of tools that enable data comparison (for a shift towards higher technology involvement). |
|
Auxiliary tools (Beta-stages, word maps, comprehensive tools) |
EPs make it possible to directly discover data on the platform using tools. Data must be available in a standardized form to best apply data exploitation tools. For EPs, a crucial success factor is the extent to which additional functions and services relate directly to the semantic dimension of its data. As GPs and BPs rarely offer facilitating tools for analysis, easy usability and accessibility of the platform is mainly considered beneficial to the users. |
|
Requirements for data upload (Data size, format, (meta) data standards) |
EPs and BPs are characterized by extensive (meta) data upload requirements, limiting the data format to upload. GPs have lower (meta) data requirements, mostly allowing for any data format, while storage limitations can apply. |
|
Quality management
|
Manual curation (Extent of platform intervention, plausibility, format checks) |
There are rather strict upload requirements for EPs, ensuring higher data quality and allowing for the use of tools to analyze standardized data. BPs facilitate the standardization of domain-specific data, which is linked to the perceived data quality. Quality can be ensured by manual curation, automatic checks, and outsourcing. Manual curation does, however, become time consuming with increasing data volumes. Since BPs have particularly low FTE values, they depend on increasing automation of data upload controls, which simultaneously enforces standardization and data quality checks. One third of the EPs and 20% of the BPs interviewed only allowed data from peer-reviewed papers that enhanced the data quality. GPs face challenges regarding the data quality offered. It is particularly noticeable that GPs with no domain restriction enforce lower metadata standards, and often have no data standards at all. A DP does not necessarily have to offer high quality data if it is not the decisive criterion for the user group. |
|
Automatic checks (Scope of underlying data model) |
|
External quality control (Data accepted from peer-reviewed articles only) |
|
Trust and credibility
|
Who favors property rights and terms of use (Availability of online documents) |
The property rights favor the author when the author has free choice about the specified License, or if the License enforces citations of the data. The data user is favored when no citation of used data is needed. The property rights favor the platform when the ownership is transferred to the platform. The property rights of EPs favor the author (need to cite) or the platform itself (transferring ownership rights). In the case of GPs and BPs, property rights mostly favor the data user (choice of License) and author (need to cite). |
|
Certifications (External rewards and assessments, like “CoreTrustSeal”) |
Certifications are mostly received by EPs, followed by GPs. Interviewees mostly cite certifications as indicators of institutional-based trust. No BP in our analysis has been certified. |
|
Incentives (Motivation for data upload)
|
Recognition (By other researchers providing measures like rankings or DOIs for citations) |
A dataset becomes citable by providing a DOI. A DOI is an incentive for data suppliers, as it acknowledges the work. In our analysis, most GPs already provide DOIs and publish rankings that enable better recognition of the data supplier on their website. In comparison, not all BPs and EPs provide DOIs. |
|
Platform outreaches (Cooperation, directly addressed target groups) |
All DPs attract new users from journal recommendations. EPs specifically, frequently advertise their high functionalities. Due to scare resources, BPs reach out to their users at scientific conferences and presentations, and by publishing research articles. Users of GPs require diverse data for their analyses, and therefore cooperate with institutions such as universities and libraries. |
|
Platform disclosure (General information, statistics, publishing names) |
GPs follow a strong platform disclosure strategy by publishing data set statistics and the name of the data provider, and by offering contact possibilities, like posting it on the main page and showing statistics for individual data sets. One GP within the sample provides a long-term data availability statement. EPs disclose the least amount of information about the data (e.g., statistics) and authors (e.g., name and contact possibilities) on the platform. The BPs disclose more information regarding contact possibilities. |
|
External incentives (Open Science policies of publishers and funders) |
No differences of external incentives between DP types have been observed. All DPs benefit from external incentives, like Open Science policies of publishers and funders. |
|
Openness
|
License agreements (Using pre-defined Licenses or offering inhouse Licenses, impact of Licenses, Licenses valid for whole platform or data set online) |
The License agreements of EPs vary in scope impact. Regarding the scope, Licenses may cover the whole platform or individual data sets only. Regarding the impact, EPs show all identified License agreements. BPs mainly comprise CC BY and CC BY-NC Licenses. It follows that data suppliers and the DP are cited by researchers when reusing them. Despite the absence of a DP citation index, citations of used data promote the work of the DP in the scientific community. For GPs, it is noticeable that either the users themselves can determine the CC-License or the entire platform is subject to the CC0 License. |
|
Involvement (Of the community in processes and procedures) |
Based on available resources, different DPs perform different tasks in community involvement. GPs conduct most surveys and workshops, followed by EPs. Another form of involvement is an ambassador program in which users can actively participate. EPs also offer further services, like tutorials, and events, such as training, to their communities. When EPs publish blog articles or spread information about news and events, they also act as a social space for the scientific community, offering users a communication platform. At this point, they can use the opportunity to actively participate in the discussion, influencing processes and procedures, and respond to their communities’ needs. Most BPs occasionally offer workshops, and then only case-based ones. |
|
Services for payment (Analytical service, workshops) |
One EP offers analytical services for data payment. |
|
Financing model
|
Revenues (Licensing, memberships, fee for data deposition) |
The biggest share of DPs receives research grants, and most research projects include database funding. Funding is especially important for BPs, as their users (mostly researchers) show the least willingness to contribute financially. Most DPs have received public funding in the early development stages. GPs are often beyond the initial (public) funding phase. In some cases, organizations follow up on public funding after their expiration. This has been observed for most BPs, but is also the usual procedure for the other types’ platforms. Institutional funding often enables BPs to extend their limited resources and develop into EPs or DPs. Licensing for additional services is most common for GPs, as their users are willing to pay for additional services, like standardized data. None of the analyzed BPs use Licensing models. Membership fees are rare within the sample, as open DPs have been the focus. Yet, such models do arise with the progression of DPs. |
