Now that I know what my natural language processing project is, I should also say what the professor had mentioned in passing but which we won't be working on. I had nicknamed it the ToA Constructor, or Taxonomy of Argument Constructor.
It works like this. Say you are debating an issue, for example "The US government should adopt universal health care." As a suggested policy of sorts, there are many arguments for and against that statement. The ToA Constructor is a tool which will help you search for different arguments which attack the motion from different directions.
Looking at arguments more globally, we can say that policies can be generally separated into two parts: the means and the end. The (implied) end in this case is to ensure the health of citizens, and the means is for the government to pay for health care through taxes or other means. Each of these two areas could be further dissected to reveal different types of supports and rebuttals.
Take the means for example. We could attack the proposal for universal health care to say that it won't actually lead to better health, or that it is too expensive, or that it won't pass congress. All these are against the proposal, but in different ways. The first one is saying that the means does not lead to the end, the second is saying that the end does not justify the means, and the third is saying that the circumstances will not allow the means to take place.
A similar deconstruction could be used for the end, to say that the benefits are too low, or that it does not help the entire nation equally. I'm sure if I take the time to examine many different motions, I will find a lot more points of attack. Notice that these points do not only apply to universal health care, but actually highlight how arguments are constructed in general, and where attacks against that argument could attach themselves.
Finally, under each of these different areas of attack, there will be examples, analogies, parallels from other countries, etc. that will support that attack. My original vision of the ToA constructor is just to apply these points of attack to the specific topic, but after that it could be extended to look at examples as well.
I did some research before I found out I won't be working on this. I was surprised that this kind of taxonomy does not already exist. I didn't find a single document which talks about how arguments are constructed, and which tries to exhaustively list ways of attacking the structure of the argument. I feel like this would be a great help to any type of debate, whether at collegiate level or in actual politics. Maybe some of you readers can point me to a good book on this topic?