The NLG Group is part of the Department of Computing Science at the University of Aberdeen. It currently has about 20 members.

Research within the group is concerned with the theoretical and practical aspects of Natural Language Generation, a branch of Computational Linguistics and Artificial Intelligence. We are interested in presenting information to human users through the automatic generation of text or speech and so we place a strong emphasis on empirical work with potential users, with a view to (a) designing systems that successfully emulate human communicative behaviour, and (b) evaluating the performance of systems on real users. Topics of active research within the NLG Group include:

  • Content determination, micro-planning and realisation in Natural Language Generation. Here, we are concerned with expanding the capability of current systems to deal with real-world scenarios in which data is available in raw formats that have not been explicitly designed as inputs to NLG systems. Another research focus is intelligent text planning that meets the requirements of coherence and cohesion while avoiding ambiguity.

  • Design and implementation of computerised tools for technical support. This includes tools to support decision-making in medical environments, as well as presentation of numerical and time-series data.

  • Data-to-Text. We are very interested in using NLG technology to generate textual summaries of non-linguistic data. Currently numerical data in particular is almost always presented to users using graphical visualisations. While visualisations work well in many cases, we believe that in some contexts textual summaries are more effective ways of communicating data to human users. We also believe that the task of generating summaries of data poses many important and interesting scientific challenges for NLG.

  • Enhancement of communicative effectiveness. The investigation of creative uses of language that make text more interesting (e.g. humour), as well as the use of multimedia and diagrammatic reasoning systems.

  • User profiling and modeling. This research area focuses on the design of systems targeting specific users such as people with limited reading skills. Another challenge is the creation of adaptive systems which respond to the needs of not only individual users, but also groups.

What's new...

10 December 2009: The Joking Computer software (an improved version of the STANDUP system) is now available to the public in Glasgow Science Centre, on a dedicated kiosk with a touchscreen interface. [more]

8 October 2009: Ehud Reiter and Yaji Sripada have formed a spin-out company Data2Text, which commercialises the data-to-text technology developed within the NLG research group. [more]

5 June 2009: There was a press conference about the "How Was School" project yesterday, which gathered a lot of attention at the BBC. The story was on the news home page, and as a video clip. Find out more about the project. [more]

24 September 2008: The University of Aberdeen's first grant under the EPSRC Partnerships for Public Engagement scheme has been awarded to Dr. Judith Masthoff and Dr. Graeme Ritchie to develop the STANDUP joke-generating software into an exhibit for science centres. [more]

4 September 2008: Funding is available to join the group for a "mini-project". [more]

The NLG Group's news