New research could lower language barriers

The Statistical Multilingual Analysis for Retrieval and Translation (SMART) project funded by the European Union (EU) and led by Xerox’s European Research Centre in France, was prompted by the fact that research by the EU suggested that more than half of Europeans can only hold a conversation in their own language, and that existing document translation services do not always produce accurate results that scan grammatically very well.

'There have been lots of applications of machine learning techniques to machine translation in the past,' said Dr Craig Saunders, the project partner at the University of Southampton's School of Electronics & Computer Science (ECS). ‘The project aims to extend the more traditional methods based on log linear models, and also apply recent developments in machine learning for structured prediction which have lead to many new powerful techniques that show great potential in this area.’

Over a three-year period, SMART plans to apply modern machine learning techniques through English, French, Spanish and Slovenian to three user scenarios.

A first user scenario will focus on the work of professional translators and aims to assess the potential impact of new technologies on their productivity.

The second scenario considers the work of technicians providing support to customers over the 'phone, holding a conversation in a language different to the technical documentation available.

The third user scenario aims to enable users to access portions of the multilingual Wikipedia in languages of which they have limited command.

'This is the first time that new machine learning techniques are being used in this way,' said Dr Saunders. 'Xerox works across lots of different languages and cross language information access could be very useful in this context; the possibility of posing a query in one language and getting documents back in another is useful in a wide variety of applications’.

The other project partners on SMART are: Amebis d.o.o., Celer Soluciones S.L., Jozef Stefan Institute, National Research Council Canada, University of Bristol, University of Helsinki, Università degli Studi di Milano and University College London.