The eduOER metadata aggregation workflow.
This section describes the main steps from which the eduOER metadata aggregation workflow consists of.
- Harvesting: The harvester mechanism used, gathers metadata from data providers using either OAI-PMH or RSS protocols.
- Transformation: The Transformation mechanism transforms harvested metadata records from their native XML Schema to the respective eduOER Application Profile (some records might already be compliant).It takes as input transformation rules defined in XSLT files(one XSLT file for every data provider).
- Identification: The identification mechanism identifies the metadata records and the learning objects they describe in order to accomplish consistency at OER repository level. An important precondition for this to work is the metadata to have information related to the learning object(e.g a URL) and the metadata its self.
- Validation: During this step the transformed metadata records are validated against OER application profile. All rules(vocabulary restrictions,mandatory elements etc) defined in the application profile should be followed by metadata records else they are considered invalid.
- Filtering: The filtering mechanism is used to check whether the mandatory elements not only exist but also contain text.
- Language Detection: The language detection mechanism is used to detect in a probabilistic way the language of specific elements(keywords,title,description etc). It could also be considered as a second filtering step since metadata records that contain elements with text that cannot be language detected are filtered out. The text that cannot be lang detected contain symbols like dashes,dots,hashes,underscores etc
- Link Checking: The link checking mechanism is used to check the learning objects' URLs defined inside metadata records.
- Metadata analysis: During this step,the final to be exposed metadata are statistically analyzed on an element based approach.