System Framework

Goals

SF's goals can be described in terms of two overall objectives: an integration objective and a research objective.

The integration objective is to produce a system services architecture and a system design methodology that supports the integration of linguistic technologies, localisation workflow and digital content management. This will enable rapid, iterative and instrumented integration of industrial software and academic research prototypes and support their evaluation through provision of: a software integration platform based on open standards, guidelines and tools for developing workflows and applications using this platform, and methods for iterative prototyping and user studies.

The research objective is to develop novel interaction design techniques and system support to the development of speech and language enabled applications. This will be achieved through the investigation and evaluation of: instrumentation of localisation processes to monitor process quality improvement; human factors assessment in localisation processes and ILT-based systems in general; analysis of contextual and culture-specific factors that affect the design of speech and language enabled systems; assessment of modality combinations and their impact on usability of multi-modal systems; and management of the quality control process by stakeholder communities.

Methodology

Although the interaction design, human factors and software engineering issues investigated in this track arise in many application contexts, there has not been a concerted attempt to investigate them in the context of emerging language and localisation technologies. The scale of the application domains we target and the novelty of the technologies which defines our design and integration spaces can therefore be regarded as the fundamental research barriers to be overcome by the Systems Framework track.

From a systems architecture and integration perspective, the methodologies employed in order to address the needs of CNGL involve enabling rapid, iterative and instrumented integration and evaluation of industrial software and academic research prototypes and supporting their evaluation through provision of:

  • A Software Integration Platform based on open standards, specially for web services (WSDL and BPEL) and localisation data (XLIFF, TMX, TBX);
  • Guidelines and tools for developing workflows and applications using this platform.

As the project develops towards more exploratory research, SF will develop novel service-oriented systems support to progressively manage the quality of Next Generation Localisation applications that are composed of language technology, digital content management and localisation workflow management services. This will be achieved through the investigation and evaluation of system support for:

  • Monitoring localisation services to deliver atomic and composite service quality management;
  • Integrating human quality assessment by content consumers into localisation and digital content management services;
  • Management of the quality control process by stakeholder communities.

From a design and human-factors perspective, SF has adopted methodologies ranging from ethnomethodologically-informed ethnography for the study of the work of translators and post-editors in situ to experimental methods for evaluation and design involving novel modality and technology combinations in expl
oratory scenarios.

Industry Engagement

Collaboration with DNP and NII, facilitated the project on perception of language output style (in the case of referential expressions) and other factors by English and Japanese users in virtual environments. SF has maintained frequent contact with DNP with a view to identifying areas for further cooperation. The work reported by Breitfuss et al. (2009) was done in collaboration with CNGL industrial partners DNP and the NII, in Tokyo. DNP was also actively involved in collaborating with Ielka van der Sluis and Saturnino Luz, having allocated a member of their staff, Junko Nagai, to the project. Ms Nagai's input was essential to this work and she co-authored the article. As regards the area of localisation, input from Symantec and a number of their language service providers presented a better understanding of current industry practices and processes. In situ observation of practices in these workplaces provided extensive data which formed the bases for SF's work on awareness mechanisms and workflows.

From a software engineering perspective, in addition to the close integration with industrial partners required as part of the Demonstrator activities, SF has been strongly engaged with industry in: co-development (with Alchemy) in integrating the ILT1 MaTrEx MT component into the Catalyst 8.0 product, collaboration with VistaTEC and Traslán and ILT3 in identifying processes in their translation review business (where the ILT3 style-based text analytics can be applied), and collaboration with Microsoft and Symantec in defining scenarios appropriate to their business in the Bulk Localisation Workflow and Personalised Multilingual Customer Care areas. The integration of MaTrEx into Catalyst 8.0 was publicly demonstrated on the Alchemy corporate stand at the Localisation Innovation Showcase held at DCU on 16 October 2009 and will form the basis of further investigations into the impact of confidence scores on in-context post-editing.

All comments are submitted to the feedback forum in the members area.