Music Information Retrieval Using Similarity Based Relevance Ranking Techniques

Abstract

The purpose of this proposed study activity is to construct a system for the job of automatically assessing the relevance of music datasets, which will be used in future work. Determine item similarity is an important job in a recommender system since it determines if two items are similar. Participants' systems must provide a list of suggested music that may be added to a given playlist based on a set of playlist characteristics, {which will work along with the algorithms designed to provide other similar songs. Specifically, in this study, the challenges of detecting music similarity only on the basis of song information and tags given by users have been addressed. The proposed technique has been tested using a variety of machine learning algorithms to see how well it performs. tf-idf} and Word2Vec are the methods used to model the dataset and generate feature vectors. It has also been found we that machine learning techniques, including Collaborative Filtering, KNN, Frequent Pattern Growth, and Matrix Factorization, have a greater influence on relevance ranking than traditional methods.

Beyond the Trends: Evolution and Future Directions in Music Recommender Systems Research

“…Two primary approaches are employed in determining similarity within music recommender systems item-oriented and user-oriented methods. Item-oriented methods focus on gauging the relevance between items, such as songs, based on their intrinsic characteristics or features [67]. Conversely, user-oriented methods estimate relevance between users by analyzing their behaviors and preferences, offering recommendations aligned with similar users' musical tastes [68].…”

“…Simultaneously, the spotlight is on incorporating personal characteristics into the recommendation process. This involves a comprehensive analysis of user attributes, demographics, and psychographics to tailor recommendations based on individual traits, adding sophistication to the user-centric recommendation experience [67].…”

Major-minorness in Tonal music -- Evaluation of Relative Mode Estimation using Expert Ratings and Audio-Based Key-finding Principles

“…However, if its goal is to define a given feature in a way that captures the musical properties in a manner pertinent to our understanding of music's perception and cognition, then a continuous treatment may in fact be more useful. Measure of relative mode implemented in MIR toolbox (Lartillot, Toiviainen, & Eerola, 2008) -a MATLAB toolkit for extracting musical features from audio -have been used to build models of emotion and mood recognition (Lin, Yang, & Chen, 2011), inform music recommendation systems such as Spotify (Vasu & Choudhary, 2022), and identify the relative "ground truth" for musical features crucial for emotion expression (Beveridge & Knox, 2009). Tools such as MIR toolbox offer the ability to automatically estimate where stimuli fall on the major/minor spectrum.…”