Misc,

D3.1.2 - Linguistic Linked Data Reference Architecture – Phase II

, , , , , , , , , , , , , , , and .
(October 2015)

Abstract

This deliverable covers the definition of a generic reference model and architecture that includes common and frequent tasks for multilingual and multimedia content analytics that require data-intensive NLP services and. The deliverable is an update of D3.1.1. Phase I and takes into account the reviewers’ comments, the discussions conducted within the Linked Data for Language Technologies (LD4LT) community group as well as the lessons learned from implementing different layers and components of the reference architecture. Major updates include the identifications of seven layers of the LIDER Reference Architecture, the inclusion of a vertical example and several implementations, as well as the refinement and an extended description of LingHub -- the implementation of the LIDER observatory for language resources --, including its evaluation. For a more concise document, we have removed state of the art analysis and description of the creation process, which can still be found in D3.1.1 Phase I1. Instead we focus on describing the results here, i.e. the LIDER Reference Architecture and its existing and future platforms, featuring: - LIDER Reference Architecture layer cake: A layer cake model that describes the necessary elements and their interplay to meet the above four mentioned requirements. - Guidelines: A set of guidelines that defines how datasets should be published to ensure their discovery, reuse etc. These are orthogonal to the LRA and have been described in D2.1.22 as part of Work Package (WP) 2. - LingHub: a data discovery service implementation - Component Model and Architectural Patterns: A set of architectural recipes that describe how frequently needed components should be implemented.

Tags

Users

  • @magarcia

Comments and Reviews