Metaphorical Meanings for Artificial Agents

The ability to understand analogy and metaphor is fundamental to human intelligence and creativity. Modelling analogy and metaphor effectively will enable us to model human language, contributing to technological advancements that can be integrated with the kind of artefact that humans use on an everyday basis, such as mobile phones or wearable tech. My research combines methods from computational linguistics, category theory, and conceptual spaces to develop a model of metaphorical meaning. The research will build on the compositional distributional model of Coecke et al., (2010), integrating this model with conceptual spaces theory (Gärdenfors, 2014). A framework for understanding metaphor will be developed, and the model as a whole will be implemented within a cognitive architecture. Importantly, this architecture allows input from different senses, which is essential in dealing with analogy. Finally, the model will be implemented within a community of interacting agents. This model will be analysed to provide insight into how new concepts spread through a community of agents. This important phase will ensure that the system developed is able to learn novel ideas by interaction with other artificial intelligence (AI) systems and with humans. The impact of my research will be to improve human interaction with computers in everyday life. I am best placed to carry out this research because of my interest and background in human concept use, evolutionary and adaptive systems, and corpus semantics. The unique combination of expertise in formal semantics, computational semantics, quantum theory and category theory within the Institute of Logic, Language and Computation (ILLC) will complement my skills perfectly and ensure the success of this project.