# CSE 464: MCQCorrector — Data

来源：https://www.kaggle.com/competitions/cse-464-mcqcorrector/data

采集时间：2026-10-19T07:33:36Z

File descriptions
train.
csv
- the training set containing information about training data.
test.
csv
- the test set file names to be corrected.
Sample Submission.csv
- a sample submission file in the correct format.
ModelAnswer.pdf
- contains the correct answer for the 45 questions in the exam.
test.zip
- the test answer sheets files to be corrected with unknown marks, you'll be evaluated based on your score on this.
train.zip
- the actual answer sheets files with known marks that you'll use to build your pipeline and make sure it's working.
Data fields (CSV fields interpretation)
FileName
- unique file name for the answer sheet located at either train/test folders.
Mark
- the correct mark for the scanned answer sheet.
BenchMark
- our ASU algorithm correction result to compare yourself with, whenever we have any uncertainty we mark an image to be revised manually, this field will be usually less than or equal the actual mark, as our algorithm ignores some questions for manual revision and only corrects questions that it's 100% sure it knows how the student answered it.
CornerCases
- this's a column to help you get the best result, it contains a list of question numbers that need to be revised when our benchmark algorithm doesn't know with 100% accuracy if the question isn't answered, or if the question have been answered with multiple answers, or if the student didn't fill the answer sheet correctly, so it needs manual revision, you're required to handle these corner cases with your algorithm, mainly this is the hardest part, you can use this column to filter hard cases to work on separately.
StudentID
- student ID read from the answer sheet.