The Digital Ecosystem Model
We are pleased to introduce the Digital Ecosystem Model (DEM), which has been developed by project partner University of Goettingen in the framework of PERICLES’ integrated approach to model-driven preservation. The DEM is based on the Linked Resource Model (LRM) also developed within the PERICLES project. In contrast to the very high level of abstraction of the LRM, the DEM represents a middle layer ontology on the way to domain specific ontologies.
The DEM is an ontology for modelling entities, dependencies and interactions of Digital Ecosystems (DE) and provides a representation of their complex relations. It uses the Web Ontology Language (OWL) which is an open standard and a wide range of tools and reasoning services are available. The concept behind the model understands Digital Ecosystems (DEs) as interwoven evolving networks of digital objects, policies, processes, technical services and user communities within constantly changing and interacting environments. DEs are characterized by having many dependencies and complex structures. Therefore it can be difficult to describe and maintain them, introduce change in a controlled way and identify and solve resulting problems. In this context the DEM provides a means to handle these difficulties and aims to improve the management and reuse of digital objects.
Overview: main aspects and functionalities
The DEM supports the description of entities and their environments not only for preservation purposes, but also for the management of environmental drift in a wider range of domains. Planned or unplanned changes to a DE represented via the DEM can be simulated to analyse and mitigate risks, and to assure the quality of the DE architecture. Change within Digital Ecosystems can occur at different levels and is not limited to technical infrastructures, dependencies or processes and workflows. Change can also occur in expectations, requirements, and background knowledge of user communities. It can be caused by merging communities or through the introduction of new policies or by the exchange of digital objects with other communities. A DE can be affected by changes in larger social or cultural contexts such as in laws, disciplines or cultural norms.
The model allows us to analyse a variety of different aspects of a Digital Ecosystem and it can be used:
- to describe the environment of entities across organisations and for ranking and weighting of entities
- to capture the state of a system and to reveal a single point of failures
- to simulate and manage change
- to analyse provenance and to trace the model evolution (history) of a DE as well as for bridging old and new concepts
- for detection of policy violations
- to record arbitrary semantic information about objects (e.g. to assign entities to a scenario and ranking the importance of the dependencies for this scenario)
- as a working basis for other tools, e.g. by consulting the DEM as an information source and automatic updating of the model in a sheer curation scenario and automated validation of the model
Design and architecture
The design of the DEM is domain agnostic, but can be extended by domain specific ontologies to support various use cases (e.g. the digital media and art domain ontologies developed in PERICLES). The DEM is built with a modular structure which allows us to reduce the overall complexity and to give more detailed descriptions of focus areas without overloading the model with unneeded entities and relations. The DEM provides a core model which is extendable by using the more granular sub-models. The following figure shows an overview over all models, which will be briefly explained in the following paragraphs. Note, that this description and figure below represent only an excerpt of available entities and relations.
DEM Core Model
The DEM Core Model defines base entities and basic principles for the overall model, which are inherited by the entities of the submodels. Hence, the DEM Core Model represents the mandatory part within the DEM. The DEM Core Model consists of five main entities: User Community, Policy, Technical Service, Process and Digital Object, which are at least necessary to define a DE. Those entities are quite abstract and further details can be specified by using the corresponding sub-models.
DEM Policy Model:
The DEM Policy Model allows setting policies as central drivers of an infrastructure and enables the creation of a policy driven DEM instance. Policies can be described as aggregations of other policies and they can be linked to processes which implement the policies. They can represent legal and organisational requirements as well as aspirational aims and directions of an organisation. More details can be modelled concerning the fulfilment level (Requirement Level), the implementation (Implementation State) or the type of policies (Policy Type).
Furthermore quality assurance elements (QA) can be included to ensure the correct application of policies. In this context the model can be used as a policy verification tool. For expressing preservation policies the model uses the SCAPE Policy Framework and the classifications of the SCAPE Guidance Policy Elements.
DEM Process Model
The DEM Process Model describes the surroundings of processes like roles, input and output relations and associated technical services where they can run on and which agent is part of the process and which entity is processed in which way. It does not model the process flow itself, a corresponding implementation (e.g. BPMN) can be linked.
The DEM Process Model offers further a semi or fully automated change management method by the process triple Process for Model Validation, Process for Entity Validation and Process for Transformation. The first process filters specified entities and passes them to the second one, where they are validated against predefined criteria. Finally, the Process for Transformation can execute corrective actions in case of a failure output from the validation process.
DEM Infrastructure Model
The DEM Infrastructure Model covers all technical infrastructure components, such as soft- and hardware, which are defined as agents, as well as technical interfaces. Infrastructure components are defined as aggregated resources consisting of several technical services. With the inclusion of policies from the DEM Policy Model it is possible to model access rights. Also maintenance responsibilities, property rights and technical requirements can be expressed.
Additionally the Infrastructure model can be used for better planning of infrastructure changes and for the management of technical components. It can help in identifying single point of failures, necessary updates or scarcity of resources.
DEM Analysis Model
The DEM Analysis Model provides a means to add supplementary information as a layer on top of DEM instances. It enables the application of arbitrary graph based analysis algorithms. The model allows the introduction of weights and annotations via annotation entities. Weighted Edges can be added to relations between entities to record meta-information about an entity.
Annotations can be formal or informal and can be further specified e.g. to define whether they depict an error or warning. A significance value can be added to entities for specific purposes or scenarios (e.g. the operating system is not significant for the Java programming language). It is possible to annotate dependencies with rules to constrain their appliance in relation to specific purposes and scenarios (e.g. a printer is only relevant in some activities). The Scenario entity allows using one DEM instance for multiple use cases by enabling a scenario specific view on a subset of entities and procedures. The Semantic Drift entity relates entities with semantic similarity or expresses drifts between different version of entities from actual and previous instances of a DEM.
DEM Modelling Strategies
The modelling of DEs is always purpose driven. Depending on the use case and purpose the modelling strategy and granularity can be very different, but can be associated to a top-down and bottom-up approach.
The top-down modelling starts by identifying the main building blocks and relate them to each other. This results in an abstract view of the DE. Details are only modelled when necessary for the scenario.
The bottom up modelling strategy starts vice versa with very specific details, like the configuration of a computer and what tasks are executed by which software. In a second step these details are connected to more general descriptions.
Both strategies can be used in parallel as the figure below illustrates, and the results can be merged into a single model.
Availability of the Digital Ecosystem Model (DEM) and the EcoBuilder tool
The DEM ontology is written in OWL and available as RDF/XML and Turtle format. For simplifying the creation of DEM instances the EcoBuilder tool has been developed, but the use is not mandatory.
The GUI allows to create Digital Ecosystem Models for scenario experts without ontology knowledge. It provides also an API for software developers facilitating the integration of DEMs into workflows. The EcoBuilder tool creates RDF/XML and Turtle as output format.
The DEM Ontology is available on GitHub at https://github.com/pericles-project/DEM
The EcoBuilder is released on GitHub at https://github.com/pericles-project/EcoBuilder. (Both are licensed as open source under Apache 2.0.)
For further information and more detailed description please see:
Johannes Biermann, Anna Eggers, Fabio Corubolo, and Simon Waddington. 2016. An ontology supporting planning, analysis, and simulation of evolving digital ecosystems. In Proceedings of the 8th International Conference on Management of Digital EcoSystems (MEDES). ACM, New York, NY, USA, 26-33. DOI: https://doi.org/10.1145/3012071.3012081
For a detailed description of the EcoBuilder tool please see deliverable D5.3 (p.44-49).
You can also find out more about the DEM and EcoBuilder in this video.