How larnhold/madmp-assessment works
Main components
• DMP Loader – Responsible for loading DMPs and turning them into a normalized form (found in the case-study and sdk packages).
• DMP Harvester Service – Provides access to DMPs and their resolved, contextual information (implemented in the core module).
• Data Store – Offers centralized access to store artifacts (present in both core and case-study implementations).
• DMP Indicator Service – The coordinator that processes DMP evaluations and generates quality indicators (core components).
• Context Loader – Retrieves external contextual data referenced in a DMP (available in the case-study package with several plugins like OpenAire, FAIRSharing, re3Data).
• Evaluator – Implements quality measurement logic for various dimensions (the evaluators are present in the case-study package again with separate evaluators such as completeness, compliance, feasibility, quality-of-actions, etc.).
• SDK Module – This module defines the common interfaces and data structures to bridge the core and case-study modules. It ensures that changes in one module do not break the other’s expected interfaces.
• External Services – Notably, the framework integrates with systems like the F-UJI FAIR evaluator (accessible via Docker Compose).
Module boundaries
• Core Module: Contains generic services (harvester, indicator, plugin loader) and configuration that is independent of the specific use case.
• Case-Study Module: Contains use case–driven implementations of loaders, evaluators, and data store configurations. It is adapted to DCS (and Science Europe) and uses multiple evaluators.
• SDK Module: Provides connectivity via common API definitions and shared models between core and case-study.
Relationships and interactions
• The DMP Loader (in the case-study module) receives raw DMP data from external or internal sources. This module may interact with the Context Loader to resolve referenced entities.
• The DMP Harvester Service (in core) gathers and aggregates both the DMP content provided by the loader and external contextual data.
• The Evaluators (in case-study) process the normalized DMPs and compute quality metrics. Their results are published back into the system.
• The DMP Indicator Service (in core) coordinates the process and stores the output in a Data Store (again, implemented in a use-case–specific way in case-study).
• Communication between core and case-study components is streamlined via the SDK interfaces and data structures.
Architectural patterns and design principles
• Modularity: Separation of concerns is achieved by dividing the system into clearly defined modules (core, case-study, sdk). This eases maintenance and enhances clarity.
• Plugin Architecture: The evaluator components, context loaders, and DMP loaders appear to use a plugin-based approach (e.g., use of PluginLoader classes) that allow flexible addition of new evaluators or data providers.
• Client-Server Model: The application is a Spring Boot service with API endpoints (e.g., Swagger is provided).
• Integration via standardized interfaces (SDK), allowing decoupling between core functionality and specific implementations.