Aberdeen has a long tradition of work in Natural Language Generation (NLG), getting computer programs to produce text in an ordinary language such as English, starting from information not expressed in language, such as tables of numerical data or a collection of entries in a database. In recent years, the NLG group has branched out into various other areas of Computational Linguistics. One example is the type of process where the generation process starts from information stated in language, and the aim is to re-phrase the text, for example to make it more readable. Other examples are computational humour (an area of Creative Computing), the analysis of sentiment in text, and the mining of logical arguments from text (related to the theory of Formal Argumentation). We’re also investigating the kinds of information that Recommender Systems should express when they suggest products to prospective buyers.
In recognition of the diversification that our group has undergone, we are starting to be known as the Computational Linguistics in AberdeeN (CLAN) research group. Our group has explored a wide range of practical uses of NLG, such as writing brief weather forecasts and summarising medical data, and this work has led to the spin-out company Arria. We are also interested in theoretical issues, such as the use of algorithms for modelling human language use, and in this area we collaborate actively with researchers in psycholinguistics in Aberdeen and elsewhere.
Recurring themes in our work include:
- controlled experiments with human users
- the study of practical applications with human users
- looking at various uses of text, not just the stating of factual information
- considering the kinds of effects that text can have on readers
- the expression of quantitative information (including “big data”)
- the generation of text that is easy to read and understand
- the study of referring expressions
The CLAN research group includes one of the largest concentrations of NLG researchers in the world, and is very active in the wider international research community. The group has weekly meetings where topical issues in and around Computational Linguistics are discussed.