Design Ideas for Automated Deduction / Automated Theorem Validation?

Question

I am the second junior student who will soon begin my dissertation, and I have an interest in automatic conclusions and automatic theoretical and theoretical methods. As in, I would like to promote art in some way (I don't mean it is pretentious, but I want to do something productive). I have Googled quite far and wide, and so far there have been few promising ideas. There are several pages of ideas for student projects, but most of them seem either terribly outdated or too advanced (I originally intended to try to synthesize postmodern thought (hahaha) and abstract its logical content, build a complete and consistent model (if possible, of course), and try to automate it by grounding the specified model, as far as possible, in a non-standard logic a la these . My advisor believed that too postmodern thought credit (some time ago I rep RNO realized postmodernism generator in Haskell with Parsec, so part of it was because of the idea), I tempt to agree.)

So yes. Does anyone have any ideas? I apologize if there is any obvious gap in my approach / if I have not completed my homework (and if so, please tell me!), But to a large extent I don’t even know where to start, and thanks for that you read all this.

Answer 1

Recognizing the front that I am not particularly well informed about the field, I was interested to learn about the usefulness (and existence) - due to the lack of a better term - an “unreliable" testing assistant, in particular in relation to problems that are usually unsolvable or insoluble. Systems based on heuristics or statistical analysis that make “guesses” and stumble, trying to get everything to work, or things along these lines. The idea, of course, would be to throw problems at it that are not amenable to more methodological methods based on exhaustive search or strict retention, possibly in bits and pieces as a component of an interactive assistant.

On the other hand, I don’t know to what extent ideas in these areas have already been studied, and I’m not sure that this is even a viable technique in the first place (if someone else knows more, be interested to hear it).

Answer 2

Some negative advice. I would avoid focusing on logical logic or first-order logic. These are areas that have been studied quite intensively, so much so that the limits of what is possible are well understood. (There are even competitions to do theoretical theorists FOL.)

You may be able to make more progress by looking at official language descriptions such as ontologies. I believe (but can be corrected) that the boundary between what is acceptable, especially with the help of an AI assistant, and what is described in the full OWL, has not yet been fully explored. Another thing ontology you might want to look at is ways to support more automated ontology creation in certain areas, although this may be too much work for sub-PhD theses.