SCS Undergraduate Thesis Topics

Aaron Snook Manuel Blum Word Problem Solving using Sequence Inference

Currently, word problems can be understood by humans but cannot be easily understood by computers. This project proposes using sequence inference as a bridge to this problem. Instead of attempting Natural Language Processing, which is difficult, a human observes a parameter in the problem, and records the answers to the question for small values of the parameter. This is often easy for the human to do but nonetheless provides an effective description of the problem to the computer. The computer extrapolates the sequence given to provide insights about the problem, which it reports back to the human, allowing the computer to participate in informal problem-solving with a human.

Close this window