Using graph databases in source code plagiarism detection

Modern plagiarism detection tools calculate the percentage of the similarity between two given source code files. In academia, the process of checking for potential plagiarized students solutions can be challenging in terms of resources due to the large number of combinations between many students. In such conditions, the reliability of plagiarism detection tools may be put to risk. Every plagiarism detection tool produces a similarity report as files containing the results of the analysis for each pair of analyzed source code files. While such a report is useful for a one-time checking, sometimes it is needed to store the result data for future use. In our previous work, the results were stored in a relational database and a list of relevant queries was defined for meaningful analysis. Nevertheless, the large number of pair-wise impacts the storage and query execution speeds. In this paper, we present a solution to this problem by importing the similarity analysis data into a graph database and evaluate the difference in the query execution speed between a graph and a relational database.