Fuzzy vectorial-based similarity detection of music plagiarism
Gianluca Zaccagnino,R. Prisco,Delfina Malandrino,Rocco Zaccagnino
DOI: https://doi.org/10.1109/FUZZ-IEEE.2017.8015655
2017-07-01
Abstract:Plagiarism, i.e., copying the work of others and trying to pass it off as one own, is a debated topic in different fields. In particular, in music field, the plagiarism is a controversial and debated phenomenon that has to do with the huge amount of money that music is able to generate. However, the existing mechanisms for plagiarism detection mainly apply superficial and brute-force string matching techniques. Such well-known metrics, widely used to discover similarities in text documents, cannot work well in discovering similarities in music compositions. Despite the wide-spread belief that few notes in common between two songs is enough to decide whether a plagiarism exists, the analysis of similarities is a very complex process. In this work, we provide novel perspectives in the field of automatic music plagiarism detection, and specifically, we propose an approach based on a fuzzy vectorial-based similarity. Given a suspicious melody, our approach envisions three steps: (1) its transformation in a vectorial representation, (2) retrieving of a list of similar melodies, (3) analysis and comparison with this subset of associated similar scores by using a fuzzy degree of similarity, that varies in a range between 0 for melodies that are fully musically different, and 1 for identical melodies. To assess the effectiveness of our system we performed tests on a large dataset of ascertained plagiarisms. Results show that it is able to reach an accuracy of 93%.
Computer Science,Mathematics