Updated 4 December 2008

David Bree

Professor David S. Brée 

 Institute for Scientific Interchange,
Torino, Italy

Emeritus Professor of Artificial Intelligence
School of Computer Science, University of Manchester, UK

email: bree AT cs.man.ac.uk


David S. Brée is currently Senior Researcher at the Institute for Scientific Interchange, Turin, Italy. He is also Emeritus Professor of Artificial Intelligence at the School of Computer Science at Manchester University, UK.

He publishes on modelling of financial markets and the semantics of natural language. His present projects include models of financial crashes and the construction and enumeration of magic squares. 

Research interests

  1. Financial markets: Financial markets play an important role in our lives, in particular when they crash. Being able to predict or understand the causes of such crashes is an interesting topic. I have recently had two PhD students in this area. One, Rob Woolfson, has built a workbench for testing the claims of practitioners who propose a strategy for price prediction - no strategy so far tested is able to outperform the market in modern times. The other, Gilles Daniel, built an agent-based model of an open book double auction financial market.  He is able to reproduce most of the stylised facts of such markets, e.g. the distribution of order arrivals, transactions and prices changes. I am currently testing a well-known model of price fluctuations preceding crashes, to see whether or not it holds up across different markets and different periods. 
  2. Magic squares: their construction and enumeration. The construction of magic squares has fascinated laymen and mathematicians alike for centuries. However there still does not exist a method for constructing all of these squares. A method for constructing a sub set of magic squares, called most-perfect squares, originally developed by Dame Kathleen Ollerenshaw, was refined by the two of us.  We can also enumerate them, for any n. I have recently developed a method for constructing and enumerating all regular pandiagonal squares.
  3. Semantics of time The way we express temporal relations in natural language varies considerably between languages. Tense and aspect have been much studied by linguists. I am particularly interested in a third method: prepositions and conjunctions. By analysing grammars of various languages it is possible to get a superficial account of the use of these temporal prepositions. However, this is never complete nor accurate. With the advent of text corpora it has been possible to study the way prepositions are actually used to indicate temporal relations. I have developed a framework for analysing such use in English. It has also been applied to Dutch and German, and, to a limited degree, to Chinese and Japanese. I am completing an analysis of the use of temporal prepositions in the old Brown corpus of American English.
  4. Event recognition People have a wonderful ability to recognise an event from a sequence of sounds, e.g. a bird's song, a musical tune or a telephone bell, even after only a few occasions. I have been interested in developing a system that will learn to recongise a temporal event from a sequence of sub events. In particular, given some model of what to expect, to adjust the temporal parameters in the model so that temporal events can be reliably distinguished. Together with my student Dr. George Paliouras, we created a system for recognising the song of the hump back whale from a very limited number of recordings.

My CV.