The problem AI Evaluator solves

Resolves the hectic task of checking papers for the teachers. The user(teacher) just uploads the answer sheet in the form of a pdf and the output consists of marks attained by the students for each question and also displays the whole score on the basis of keywords (or words synonymous to it) included in the answer.

Challenges we ran into

Initially there were discrepancies in the evaluation of the marks but by using efficient and innovative algorithms, we have eliminated that issue.
Since, we were using keyword based evaluation system, so it was difficult to recognize words similar to the keyword. We have used Natural Language Toolkit (NLTK) to check for presence of synonymous words in the answers.
There were contradictions between datasets provided by the user (teacher) and NLTK on the basis of keyword priority, so we have intersected the datasets intelligently to solve the issue.
Some advanced modifications in the code fixed the initial probem of integrating the backend with the frontend.
Since, we were using the 'custom tkinter' (which is a newer version of tkinter), there were not much resources to rely on, but enough for us to come up with a stable and well build frontend for the project.