Model-based Natural Language Generation
The use of sensors, both in conventional industrial settings and in the modern ‘internet of things’ settings, generates large quantities of data – so called Big Data. In view of the data overgrowth, we need automated technologies to assist humans in comprehending it. There are several technologies that operate in this space – data mining, expert systems, Model-based Systems and Qualitative Reasoning (MBS&QR), natural language generation (NLG) and information visualization (InfoVis). While data mining/machine learning, MBS&QR and expert systems focus on the interpretation of data, NLG and InfoVis focus on information presentation. In reality a combination of these technologies is most likely to offer a practical solution to the problem of data/information overload. In particular, model-based data interpretation technologies provide the means to develop deep insights into the input data while information presentation technologies provide means to present these deep insights to human users.
At Aberdeen we have made significant contributions to this in two strands: ab initio machine learning of qualitative models of dynamic systems; and the use of qualitative models as the basis for deeper causal NLG (model-based natural language generation - MBNLG).
In this project we will combine and extend these strands; in particular to address the following questions:
1. whether and to what extent it is possible to learn ab initio, fuzzy, qualitative and semi-quantitative models
2. do these models provide adequate explanatory power that will enable MBNLG at a broader range of abstractions.
We will achieve this by means of evolutionary or immune-inspired computation approaches to perform effective search in large-scale model spaces. We will assess the models thus learnt by using them in test-bed NLG system developed using existing Aberdeen Data-to-Text NLG technology to determine their explanatory power.
The successful applicant should have, or expect to have, an Honours Degree at 2.1 or above (or equivalent) in Computing Science or related disciplines. Knowledge - Essential: evolutionary computing; machine learning basics; programming in Java, Python, or Ruby. Desirable: modelling and simulation; natural language generation; artificial immune system; Model-based Reasoning;
The other supervisor on this project is Dr Y Sripada, Computing Science
There is no funding attached to this project, it is for self-funded students only.
Formal applications can be completed online: http://www.abdn.ac.uk/postgraduate/apply. You should apply for PhD in Computing Science, to ensure that your application is passed to the correct College for processing. Please ensure that you quote the project title and supervisor on the application form.
Informal inquiries can be made to Professor G Coghill, ([email protected]) with a copy of your curriculum vitae and cover letter. All general enquiries should be directed to the Graduate School Admissions Unit ([email protected]).