The problem Modern Education: 'Ctrl+C' + 'Ctrl+V' solves

Plagiarism is on its rise especially in today's scenario where everything is operated and evaluated digitally. Students and teachers all over the world are facing this problem. The main reason we took topic is because of its effect on the learning and the integrity of student, researchers as well as teachers. The more we observe the clearer it becomes that when one starts using plagiarism he/she gets into the habit of passing someone's work as their own, this stops the learning process of the individual altogether, as that person gets rewarded for doing nothing at all. This also forces many people to switch to copy-paste habit as their hard work does not get appreciated meanwhile the ones' who cheated get more rewarded. Our motivation is to survey and understand the reasons and types of such practices and try to come up with a better solution.

Most of the available and proposed models are uni-dimensional and thus treat one aspect of data at a time that is either text or code. We are proposing a website to deal with online as well as peer-to-peer plagiarism. This website or prototype will not only be able to check for the textual similarity of articles but also code similarity (as implemented in MOSS checkers) at the same time. We are proposing to use python libraries for textual similarity detection (online as well as peer-to-peer) and implementation of available resources for code similarity detection.

Challenges we ran into

The existing challenges of present detectors are:

1. Available detectors are uni-dimensional.
2. Many free plagiarism checkers only check work against websites – not against books, journals or papers previously submitted by other students.
3. Most plagiarism checkers are only able to detect “direct plagiarism”, or instances where the sentences are exactly the same as in the original source.
4. Most of the code similarity get deceived if you are able to change all the variable names, the structure of the code, modify functions, change if statements to switch statements.

Our proposed model can deal with the following problems from above:

1. Multi-dimensional approach i.e Text and code parsing.
2. Peer-to-peer plagiarism detection along with online detection.
3. Percentage and doubt highlighting over the suspected region to deal with patchwork plagiarism.

The challenges we are predicting to occur in our model are:

1. Need for machine learning, to detect and segregate text from code and handle them respectively.
2. Need for website or dynamic deployment platform for better flexibility i.e UI.