Misc,
D3.1.1 - Linguistic Linked Data Reference Architecture – Phase I
(November 2014)
Abstract
This deliverable covers the definition of a reference architecture that includes common and frequent tasks that require NLP services and free, open and interoperable (FOI) resources for multilingual and multimedia content analytics. The LIDER reference architecture is based on open standards and existent and future platforms and provides two things:
A reference model that identifies those tasks where Linguistic Linked Data can support content analytics and provides a standard decomposition of those tasks into elements that cooperatively could solve those tasks together with the data flows between the elements.
A catalogue of architectural patterns that describes the types of elements that can be used in the abovementioned tasks, the types of relationships between those elements, and the constraints on how they may be used.
Section 1 and 2 describe different sources of input to this reference architecture and problems and barriers which are only partially solved. Furthermore, we put the LIDER reference architecture into context of existing work in the Linked Data area.
In the remaining sections, the consortium has collected and streamlined an extensive collection of NLP tasks and architectural patterns. This collection is used as a foundation to design a coherent reference architecture that guides linked data early adopters from the LIDER target groups of industrial stakeholder.
