Looking for input on a machine learning/ natural language processing project

Hi all,

I'm working on a project for school where I intend to design and implement a classifier, which can identify the semantic relationship of synonomy using a machine learning algorithm. This will be my first major project in this area, so I'm trying to get as much input as possible on how I will approach the problem. As I see it, there are three major obstacles that must be dealt with:

1. Choosing an appropriate representation of English Syntax.

Since finding semantic relationships must depend, at least somewhat, on complex syntactical relationships, I'd like to utilize a finite state machine to represent English grammar. That is, I'd like to represent each word with both its part of speech (as tagged by StanfordNLP's POS tagger) and its relationship to other words in the sentence, ie does it fall at the end of a noun phrase?

Does anyone know of an easily accessible comprehensive list of syntactic structures for the English language, or better of a well developed finite state representation of these structures?

1. Development of a training set

I will be using a text of my choosing, identifying each word with a variable, its POS, and syntactic relation, and identifying all other synonymous variable.

1. Use a support vector machine to separate out synonymous concepts from the text based on POS and syntactic relationships.

I will do one concept class at a time, and I am hoping to find some sort of pattern that relates POS, syntactic structure, and semantic synonomy together. I'm choosing to use support vector machines because of the ability to use the kernal trick in order to look at many combinations of variables (words).

Obviously, this is just a simple outline of what I'm attempting to do, but I would really love some help and input from the community. My main background is in biology, physics, and neuroscience and just started in computer science in the last year, so I am somewhat lacking in experience. I have quite a bit of time before the project is due, but I would like to get started as soon as possible. Thanks!