From experience, we have learned that a set of services with specific goals are necessary for enabling share and reuse. ARIADNE has been improving these services constantly. However, with the inclusion of a number of new services such as the registry, the transformation and the identification service, ARIADNE has now reached a level of maturity leading to a much greater uptake.
The Repository services allow for the management of learning objects in an open and scalable architecture. To enable stable querying, publishing, and harvesting of digital learning material, all key technologies that have been discussed above are used in ARIADNEs repository services.
The Registry service is a catalog service that provides up-to-date information on learning object repositories (LORs). It provides the information necessary for systems to be able to select the appropriate protocols such as OAI-PMH, SQI, SPI, SRU/SRW supported by a given learning object repository. The information is captured using the IMS LODE specification, that is based on the Dublin Core Collections Application Profile, complemented with ISO 2146 and IEEE LOM.The registry service facilitates interoperability between numerous learning object repositories. Besides that, creating a network of interoperable LOR registries, allows for the automatic discovery of new repositories with interesting material. All information in the ARIADNE registry is exchanged with all registries in the network, with the consequence that if content provider X adds his target to the ARIADNE registry, client tools of the other registries like the EUN and LACLO ones, can also discover the repository. The same goes for provider Y who adds his target to the LACLO registry. Therefore, it can be found by client tools of the ARIADNE and the EUN registries. Such client tools could for example be harvesters that would automatically harvest all targets from a registry.
The ARIADNE harvester uses the OAI-PMH framework for harvesting metadata instances from an OAI-PMH target and publishes them with the Simple Publishing Interface (SPI) into
- a specified component of the ARIADNE storage layer,
- any other repository that has an SPI target on top of its repository.
As an illustration, in the ASPECT eContentPlus project, the ARIADNE harvester is used to harvest metadata from thirteen ASPECT content providers. The harvester then publishes their metadata in the Learning Resource Exchange (LRE). As a result, teachers can then discover these resources via the ASPECT portal. This is the client tool they are familiar with for searching learning material for their courses. Once configured, the harvester can autonomously manage the addition, deletion and updates of metadata in a periodic and incremental way.
The validation service is available for providing validation of metadata instances against predefined application profiles, for example based on IEEE LOM. To ensure that only compliant metadata are stored in the ARIADNE repository, we use the validation service to check both the syntactic and semantic validity of the instances against the used profiles. The validation service has a modular approach, and combines different sorts of validation techniques including:
- XSD schema, mainly used for structural validation of xml metadata.
- Schematron rules, which are very flexible and are used for:
- verifying the presence of mandatory elements
- checking the presence of empty attribute fields. For example, in the langstring datatype of LOM, the language of a string should be recorded in a non-empty attribute “language”.
- enforcing inter-field dependencies, like conditional fields.
- checking for the correct use of the allowed terms in a vocabulary.
- validation of vcards present in the metadata with a separate vcard parser or validator.
The validation service is available as an online service where one single metadata record can be validated against the appropriate scheme. It is also integrated in the ARIADNE harvester for validating large sets of records. Reports are automatically generated which give an overview of all validation errors.
The transformation service converts metadata in one format, e.g Dublin Core (DC), into another format; e.g. the ARIADNE application profile in LOM. We need this transformation service due to the multiplicity of different metadata schemes that are used in various networks of learning object repositories. For example if we want our end users to be able to discover content, described with DC metadata, in our IEEE LOM driven query tool, we need to transform the DC to IEEE LOM first and store this representation in the ARIADNE repository. As there are a wide variety of standards and custom formats used, the transformation service works with a growing set of transformers. These transformers transform (i) metadata from one standard to another (like DC to LOM), (ii) from one AP to another within the same standard or (iii) they can combine both. Thus for every set of metadata the appropriate transformer needs to be selected, adapted or written if it does not exist yet.
The Identification service is used to provide persistent digital identifiers to resources in the ARIADNE infrastructure. The HANDLE system is used as the backend service to create globally unique, persistent and independent identifiers. This system allows the assignment, management and resolution of persistent identifiers in a distributed environment. The lower level API provided by the HANDLE system is connected to an independent service interface that provides the basic functionality for persistent storage: Create, Resolve, Update and Delete (CRUD) identifiers. The identifiers created by the service are compliant with the Universally Unique Identifier standard. For this purpose the Java Uuid Generator (JUG) is used.
The SamgI service is able to semi-automatically generate metadata instances. Through automatic metadata generation, by extracting relevant information from contents and contexts, it is possible to significantly remove the need to fill in electronic forms when describing learning resources with metadata. This is essential if we want to make our approach scale up and become mainstream.
The Federated Search Service relies on SQI to offer transparent search to both ARIADNE and GLOBE.
The Ranking service ranks search results according to Contextualized Attention Metadata (CAM) which captures social data about the learning objects such as the number of times an object has been downloaded by an end user.
The ALOCOM service supports two processes: the disaggregation of learning objects into their components (text fragments, images, definitions, diagrams, tables, examples, audio and video sequences) as well as the automatic assembly of these components in real-world authoring tools.