Modelling creativity: using statistical natural language processing techniques to identify key components

(Joint work with Anna Jordanous, University of Kent, 2010 - present)

Creativity is a complex, multi-faceted concept encompassing a variety of related aspects, abilities, properties and behaviours. If we wish to study creativity scientifically, then a tractable and well-articulated model of how creativity behaviour emerges is required. Such a model would be of great value to researchers investigating the nature of creativity and in particular, those concerned with the evaluation of creative practice. This work adopts a unique approach to developing a suitable model of how creative behaviour emerges that is based on the words people use to talk about creativity. Using techniques from the field of statistical natural language processing, key components/attributes of creativity are identified through an analysis of a corpus of academic papers on the topic. The components provide an ontology of creativity: a set of building blocks which can be used to model and evaluate creative practice in a variety of domains.

The following lexical data has been generated as part of this work and is made freely available. For a full explanation, please see the README file:

A Unified Model of Compositional and Distributional Semantics: Theory and Applications

(EPSRC EP/I037458/1 2012 - 2015)
The notion of meaning is central to many areas of Computer Science, Artificial Intelligence (AI), Linguistics, Philosophy, and Cognitive Science. A formal account of the meaning of natural language utterances is crucial to AI, since an understanding of natural language is at the heart of much intelligent behaviour. More specifically, Natural Language Processing (NLP) requires a model of meaning for many of its tasks and applications.
There have been two main approaches to modelling the meaning of natural language in NLP. First, using Frege’s principle — the meaning of a compositional phrase is a function of the meanings of its parts and how those parts are combined — logicians have developed formal accounts of how the meaning of a sentence can be determined from how the words in a sentence are related. Second, using ideas that can be traced to the linguist Firth and the Wittgensteinian dictum of “meaning as use” the distributional approach determines the meanings of words by considering the contexts in which they appear in text. The two approaches can be roughly characterized as follows: the compositional approach is concerned with how meanings combine, but has nothing to say about the individual meanings of words; the distributional approach is concerned with word meanings, but has nothing to say about how those meanings combine.
This ambitious project is investigating ways of exploiting the strengths of the two approaches, by developing a unified model of distributional and compositional semantics. We aim to make the following fundamental contributions:
  1. advance the theoretical study of meaning in Linguistics and Computer Science;
  2. develop new meaning-sensitive approaches to NLP applications which can be robustly applied to naturally occurring text.
Our contention is that language technology based on “shallow” approaches is reaching its performance limit, and to produce the next generation of intelligent language technology requires a more sophisticated, but robust, model of meaning. Our novel combination of logical and distributional approaches promises to provide the first compositional model of phrase and sentence meaning which can be effectively and robustly applied to real-world text.

Natural Language Service Composition

(EPSRC GR/S26408/01 2003 - 2007)
The pervasive computing environment of the future will provide a wide variety of services within the network. The value of such services will be greatly enhanced if the user is able to tailor her computing environment by composing services. We propose investigate how natural language processing (NLP) techniques can help make service composition a possibility for non-techn users. We focus on the development of an interactive service composition tool that uses a NL interface. Services advert themselves to a service directory that will develop an ontology of services. This can be exploited to constrain the discourse dom in which users describe new features. The interface will guide the user to an unambiguous expression of the new feature that can mapped onto a standard component architecture of services. The project concerns the practical viability of NLP technology and evaluation is a key concern. Development will be guided by a realistic corpus of feature descriptions obtained from a pool of n technical users, and by usability trials testing the viability of the approach. The final phases of the project will involve an evaluatior the completed system.