This dissertation was an inquiry and proof of concept on using NLP techniques for context based keyword association for online advertising.
The proof of concept used the NLTK Package for Python which allowed for rapid development and prototyping of the system. The system has a slow set up as the copra are loaded into memory, however when used it is speedy, but not at production level.
The system mainly used keyword tagging with some algorithms developed in Lancaster University. However the report dives deep into many of the methods used within corpus based linguistics.
The conclusion was that yes it is possible however there would potentially need to be more meta data tagging pages before had, and that a greater usage of statistical methods rather then tagging methods may allow for greater accuracy.
Copy Right held by Daniel Kershaw