Senior Data Scientist
In my thesis, I explored how to assign keywords and phrases accurately to single documents, based upon only their content and a thesaurus reference document. The findings of the PhD were that the approach undertaken had merit, but that the method investigated of using synonyms was not significantly better than a simple count of frequently occurring words.
There were a number of publications arising from this work. One presentation was given to the British Association of Applied Linguistics conference. Two papers were presented to the third and fourth eKNOW Conference on Information Processes and Knowledge Management: both of which received a Best Paper award. One paper was published in the International Journal on Advances in Life Sciences; and one paper was published in the International Journal of Advances in Software.
In my Masters dissertation, I explored using a Learner Classifier System to train a robot to use simulated emotions to improve its ability to navigate in an unknown environment.
The robot had six emotions, three positive (appetitive) and three negative (aversive): two related to current circumstances (Happiness (P+) and Sadness (P-); proportional emotions); two related to past circumstances (Curiosity (I+) and Anger (I-); integral emotions); and two based on future/predicted circumstances (Hope (D+) and Fear (D-); derivative emotions).
| Time Period | Type | Positive | Negative |
|---|---|---|---|
| Past | Integral | Curiosity | Anger |
| Present | Proportional | Happiness | Sadness |
| Future | Derivative | Hope | Fear |
The LCS was used to learn the effective use of the emotions. I.e. for a given emotional state, what is the best action to take – forwards, backwards, turn, etc.