Music Recommendation Research Articles

Retracted] Music Recommendation Algorithm Based on Multidimensional Time‐Series Model Analysis

This paper proposes a personalized music recommendation method based on multidimensional time‐series analysis, which can improve the effect of music recommendation by using user’s midterm behavior reasonably. This method uses the theme model to express each song as the probability of belonging to several hidden themes, then models the user’s behavior as multidimensional time series, and analyzes the series so as to better predict the use of music users’ behavior preference and give reasonable recommendations. Then, a music recommendation method is proposed, which integrates the long‐term, medium‐term, and real‐time behaviors of users and considers the dynamic adjustment of the influence weight of the three behaviors so as to further improve the effect of music recommendation by adopting the advanced long short time memory (LSTM) technology. Through the implementation of the prototype system, the feasibility of the proposed method is preliminarily verified.

Retracted] Optimization of an Intelligent Music‐Playing System Based on Network Communication

Streaming media server is the core system of audio and video application in the Internet; it has a wide range of applications in music recommendation. As song libraries and users of music websites and APPs continue to increase, user interaction data are generated at an increasingly fast rate, making the shortcomings of the original offline recommendation system and the advantages of the real‐time streaming recommendation system more and more obvious. This paper describes in detail the working methods and contents of each stage of the real‐time streaming music recommendation system, including requirement analysis, overall design, implementation of each module of the system, and system testing and analysis, from a practical scenario. Moreover, this paper analyzes the current research status and deficiencies in the field of music recommendation by analyzing the user interaction data of real music websites. From the actual requirements of the system, the functional and performance goals of the system are proposed to address these deficiencies, and then the functional structure, general architecture, and database model of the system are designed, and how to interact with the server side and the client side is investigated. For the implementation of data collection and statistics module, this paper adopts Flume and Kafka to collect user behavior data and uses Spark Streaming and Redis to count music popularity trends and support efficient query. The recommendation engine module in this paper is designed and optimized using Spark to implement incremental matrix decomposition on data streams, online collaborative topic model, and improved item‐based collaborative filtering algorithm. In the system testing section, the functionality and performance of the system are tested, and the recommendation engine is tested with real datasets to show the discovered music themes and analyze the test results in detail.

Retracted] Using Factor Decomposition Machine Learning Method to Music Recommendation

The user data mining was introduced into the model construction process, and the user behavior was decomposed by analyzing various influencing factors through the factorization machine (FM) learning method. In the recommendation screening stage, the collaborative filtering recommendation is combined to screen the recommendation candidate set. The idea of user‐based collaborative filtering (CF) is used for reference to obtain music works favored by similar users. On the other hand, we learn from item‐based CF, which ensures that the candidate set covers user preference. Firstly, the user’s interest value is predicted by using dynamic interest model. Then, the common problems such as cold start and hot items processing are fully considered. The frequent pattern growth algorithm is compared with the association rule algorithm based on the collaborative filtering recommendation algorithm and the content‐based recommendation algorithm, which proves the superiority of the algorithm and its role in solving the recommendation problem after applying the recommendation. The music data in the database data conversion effectively improve the efficiency and accuracy of mining. According to the implementation of the algorithm described in this article, the accuracy of the music recommendation results used to recommend user satisfaction is proved. And the recommended music is indeed feasible and practical.

Learning Dynamic User Interest Sequence in Knowledge Graphs for Click-Through Rate Prediction

Despite that path-based and embedding-based models with knowledge graphs (KGs) achieve better recommendation performance compared with other deep learning based methods, such improvement is limited due to a lack of modeling user's dynamic interest. To address this issue, we explore a principled model to provide semantic understanding of each item in user's historical interest sequence in KGs. Specifically, we propose a multi-granularity dynamic interest sequence learning method, which is based on knowledge-enhanced path mining and interest fluctuation signal discovery, to obtain semantic-enhanced paths. Furthermore, the paths are embedded by the SEP2Vec, and merged through the proposed entropy-aware pooling layer to obtain the user preference representation, which is then used to learn dynamic user interest sequence. Experimental results on two public datasets of movie and music recommendation, and two industrial datasets of personalized local service recommendation in Alipay App have illustrated that the proposed model can achieve significantly better prediction performance compared with other known baselines.

Design of the Piano Score Recommendation Image Analysis System Based on the Big Data and Convolutional Neural Network.

In the era of big data, the problem of information overload is becoming more and more obvious. A piano music image analysis and recommendation system based on the CNN classifier and user preference is designed by using the convolutional neural network (CNN), which can realize accurate piano music recommendation for users in the big data environment. The piano music recommendation system based on the CNN is mainly composed of user modeling, music feature extraction, recommendation algorithm, and so on. In the recommendation algorithm module, the potential characteristics of music are predicted by the regression model, and the matching degree between users and music is calculated according to user preferences. Then, music that users may be interested in is generated and sorted in order to recommend new piano music to relevant users. The image analysis model contains four “convolution + pooling” layers. The classification accuracy and gradient change law of the CNN under RMSProp and Adam optimal controllers are compared. The image analysis results show that the Adam optimal controller can quickly find the direction, and the gradient decreases greatly. In addition, the accuracy of the recommendation system is 55.84%. Compared with the traditional CNN algorithm, this paper uses the convolutional neural network (CNN) to analyze and recommend piano music images according to users' preferences, which can realize more accurate piano music recommendation for users in the big data environment. Therefore, the piano music recommendation system based on the CNN has strong feature learning ability and good prediction and recommendation ability.

Leveraging Affective Hashtags for Ranking Music Recommendations

Mood and emotion play an important role when it comes to choosing musical tracks to listen to. In the field of music information retrieval and recommendation, emotion is considered contextual information that is hard to capture, albeit highly influential. In this study, we analyze the connection between users` emotional states and their musical choices. Particularly, we perform a large-scale study based on two data sets containing 560,000 and 90,000 #nowplaying tweets, respectively. We extract affective contextual information from hashtags contained in these tweets by applying an unsupervised sentiment dictionary approach. Subsequently, we utilize a state-of-the-art network embedding method to learn latent feature representations of users, tracks and hashtags. Based on both the affective information and the latent features, a set of eight ranking methods is proposed. We find that relying on a ranking approach that incorporates the latent representations of users and tracks allows for capturing a user's general musical preferences well (regardless of used hashtags or affective information). However, for capturing context-specific preferences (a more complex and personal ranking task), we find that ranking strategies that rely on affective information and that leverage hashtags as context information outperform the other ranking strategies.

Deep Multi-Objective Multi-Stakeholder Music Recommendation

Many successful recommendation approaches rely on the optimization of a single objective function and focus on predicting ratings by considering customer and product features. In this paper, we consider the multi-objective recommendation problem for several stakeholders and introduce a large-scale recommender system that aims at satisfying multiple, potentially conflicting, objectives of consumers and vendors. We propose a two-tower neural network model for music recommendation that: (a) learns each of the stakeholders' objectives in a different tower, (b) shares the latent information that was learned in each tower to predict each stakeholder's objective, and (c) aggregates the predicted objectives to generate rating-based recommendations. Specifically, we focus on the following criteria: user satisfaction objectives, such as saves, likes, and degrees of engagement with songs; and artist satisfaction objectives, such as acquiring new fans. We apply our proposed deep architecture to music recommendation and examine the performance of our model on two proprietary industrial-scale datasets provided by Spotify and on a public music listening history dataset from Last.fm. We find that our model is effective in solving the multi-objective problem and achieves the best performance trade-off for the two stakeholders (users and artists), when compared to the original single-objective model and state-of-the-art multi-objective and multi-stakeholder methods.

Music Style Classification Algorithm Based on Music Feature Extraction and Deep Neural Network

The music style classification technology can add style tags to music based on the content. When it comes to researching and implementing aspects like efficient organization, recruitment, and music resource recommendations, it is critical. Traditional music style classification methods use a wide range of acoustic characteristics. The design of characteristics necessitates musical knowledge and the characteristics of various classification tasks are not always consistent. The rapid development of neural networks and big data technology has provided a new way to better solve the problem of music‐style classification. This paper proposes a novel method based on music extraction and deep neural networks to address the problem of low accuracy in traditional methods. The music style classification algorithm extracts two types of features as classification characteristics for music styles: timbre and melody features. Because the classification method based on a convolutional neural network ignores the audio’s timing. As a result, we proposed a music classification module based on the one‐dimensional convolution of a recurring neuronal network, which we combined with single‐dimensional convolution and a two‐way, recurrent neural network. To better represent the music style properties, different weights are applied to the output. The GTZAN data set was also subjected to comparison and ablation experiments. The test results outperformed a number of other well‐known methods, and the rating performance was competitive.

Physiological State Can Help Predict the Perceived Emotion of Music: Evidence from ECG and EDA Signals

As the soul of music, emotion information is widely used in music retrieval and recommendation systems because the pursuit of emotional experience is the main motivation for music listening. In the field of music emotion recognition, computer scientists investigated computation models to automatically detect the perceived emotion of music, but this method ignores the differences between listeners. To provide users with the most accurate music emotion information, this study investigated the effects of physiological features on personalized music emotion recognition (PMER) models, which can automatically identify an individual’s perceived emotion of music. Applying machine learning methods, we formed relations among audio features, physiological features, and music emotions. First, computational modeling analysis shows that physiological features extracted from electrocardiogram and electro-dermal activity signals can predict the perception of music emotion for some individuals. Second, we compared the performance of physiological feature-based perception and feeling models and observed substantial individual differences. In addition, we found that the performance of the perception model and the feeling model is related in predicting happy, relaxed, and sad emotions. Finally, by adding physiological features to the audio-based PMER model, the prediction effect of some individuals was improved. Our work investigated the relationship between physiological state and perceived emotion of music, constructed models with practical value, and provided a reference for the optimization of PMER systems.

Addressing long tail problem in music recommendation systems

Addressing long tail problem in music recommendation systems

Musical Preference: Role of Personality and Music-Related Acoustic Features

Personality factors, typically determined by the Big Five Inventory (BFI), have been a primary method for investigating individual preferences in music. While these studies have yielded a number of insights into musical choices, weaknesses exist, owing to the methods by which music is characterized and categorized. For example, musical genre, music-preference dimensions (e.g., reflective and complex), and musical attributes (e.g., strong and mellow), reported within the literature, have arguably produced inconsistent and thus difficult to interpret results. We attempt to circumvent these inconsistencies by classifying music using objectively quantifiable acoustic features that are fundamental to Western music, such as tempo and register. Moreover, it is our contention that the link between musical preference and personality may operate primarily at the level of acoustic features and not at broader categorization levels, such as genre. This study attempts to address this issue. Ninety participants listened to and indicated preference for stimuli that were systematically manipulated by dynamics (attack rate), mode, register, and tempo. Personality was measured using the BFI, allowing for analysis of personality traits and preference for acoustic features. Results supported the link between personality and preference for certain acoustic features. Preference with respect to dynamics was related to openness and extraversion; mode to conscientiousness and extraversion; register to extraversion and neuroticism; and tempo to conscientiousness, extraversion, and neuroticism. Though significant, these associations were relatively weak; therefore, future research could expand the number of manipulated acoustic features. Specific attempts should also aim to disentangle the effects of genre versus acoustic features on musical preferences. Personality–preference relationships at the acoustic-feature level are discussed with respect to music recommender systems and other aspects of the literature.

Addressing long tail problem in music recommendation systems

Addressing long tail problem in music recommendation systems

Predicting the Preference for Sad Music: The Role of Gender, Personality, and Audio Features

The “tragedy paradox” of music, avoiding experiencing negative emotions but enjoying the sadness portrayed in music, has attracted a great deal of academic attention in recent decades. Combining experimental psychology research methods and machine learning techniques, this study (a) investigated the effects of gender and Big Five personality factors on the preference for sad music in the Chinese social environment and (b) constructed sad music preference prediction models using audio features and individual features as inputs. Statistical analysis found that males have a greater preference for sad music than females do, and that gender and the extraversion factor are involved in significant two-way interactions. The best-performing random forest regression shows a low predictive effect on the preference for sad music ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$R^{2} =0.138$ </tex-math></inline-formula> ), providing references for music recommendation systems. Finally, the importance-based model interpretation feature reveals that, in addition to the same music inputs (audio features), the perceived relaxation and happiness of music play an important role in the prediction of sad music preferences.

Personalized Music Recommendation System

The main aim of MRS is to give proper meaningful suggestion to person for specific items based on users mood and interest towards particular items, MRS has seen a boom in recent years. The two most popular recommendation algorithms: 1) Content-Based Filtering and 2) Collaborative Filtering. The Content Based method recommends music based on the user data, In Collaborative method we use rating and sharing of content between different users to recommend music. We take help to provide the music recommendation by content based method music subjective features are like the Speechiness, loudness and Acoustiness etc. are analyzed. Cold-start is the most common problem for new users. Here, most popular tracks are recommended to users to solve it.

Towards Emotion-aware Recommender Systems: an Affective Coherence Model based on Emotion-driven Behaviors

Decision making is the cognitive process of identifying and choosing alternatives based on preferences, beliefs, and degree of importance given by the decision maker to objects or actions. For instance, choosing which movie to watch is a simple, small-sized decision-making process. Recommender systems help people to make this kind of choices, usually by computing a short list of suggestions that reduces the space of possible options. These systems are strongly based on the knowledge of user preferences but, in order to fully support people, they should be grounded on a holistic view of the user behavior, that includes also how emotions, mood, and personality traits influence her choosing patterns.In this work, we investigate how to include emotional aspects in the recommendation process. We suggest that the affective state of the user, defined by a set of emotions (e.g., joy, surprise), constitutes part of choosing situation that should be taken into account when modeling user preferences.The main contribution of the paper is a general emotion-aware computational model based on affective user profiles in which each preference, such as a 5-star rating on a movie, is associated with the affective state felt by the user at the time when that preference was collected.The model estimates whether an unseen item is suitable for the current affective state of the user, by computing an affective coherence score that takes into account both the affective user profile and not-affective item features. The approach has been implemented into an Emotion-aware Music Recommender System, whose effectiveness has been assessed by performing in-vitro experiments on two benchmark datasets. The main outcome is that our system showed improved accuracy of recommendations compared to baselines which include no affective information in the recommendation model.

Personalized Music Recommendation Simulation Based on Improved Collaborative Filtering Algorithm

Collaborative filtering technology is currently the most successful and widely used technology in the recommendation system. It has achieved rapid development in theoretical research and practice. It selects information and similarity relationships based on the user’s history and collects others that are the same as the user’s hobbies. User’s evaluation information is to generate recommendations. The main research is the inadequate combination of context information and the mining of new points of interest in the context-aware recommendation process. On the basis of traditional recommendation technology, in view of the characteristics of the context information in music recommendation, a personalized and personalized music based on popularity prediction is proposed. Recommended algorithm is MRAPP (Media Recommendation Algorithm based on Popularity Prediction). The algorithm first analyzes the user’s contextual information under music recommendation and classifies and models the contextual information. The traditional content-based recommendation technology CB calculates the recommendation results and then, for the problem that content-based recommendation technology cannot recommend new points of interest for users, introduces the concept of popularity. First, we use the memory and forget function to reduce the score and then consider user attributes and product attributes to calculate similarity; secondly, we use logistic regression to train feature weights; finally, appropriate weights are used to combine user-based and item-based collaborative filtering recommendation results. Based on the above improvements, the improved collaborative filtering recommendation algorithm in this paper has greatly improved the prediction accuracy. Through theoretical proof and simulation experiments, the effectiveness of the MRAPP algorithm is demonstrated.

Structured Gaussian Process Regression of Music Mood

Modeling the music mood has wide applications in music categorization, retrieval, and recommendation systems; however, it is challenging to computationally model the affective content of music due to its subjective nature. In this work, a structured regression framework is proposed to model the valence and arousal mood dimensions of music using a single regression model at a linear computational cost. To tackle the subjectivity phenomena, a confidence-interval based estimated consensus is computed by modeling the behavior of various annotators (e.g. biased, adversarial) and is shown to perform better than using the average annotation values. For a compact feature representation of music clips, variational Bayesian inference is used to learn the Gaussian mixture model representation of acoustic features and chord-related features are used to improve the valence estimation by probing the chord progressions between chroma frames. The dimensionality of features is further reduced using an adaptive version of kernel PCA. Using an efficient implementation of twin Gaussian process for structured regression, the proposed work achieves a significant improvement in R2 for arousal and valence dimensions relative to state-of-the-art techniques on two benchmark datasets for music mood estimation.

Datafication and the push for ubiquitous listening in music streaming

This article discusses Spotify’s approach to music recommendation as datafication of listening. It discusses the hybrid types of music recommendation that Spotify presents to users. The article explores how datafication is connected to Spotify’s push for the personalization and contextualization of music recommendations based on a combination of the cultural knowledge found in editorial curation and the potential for large-scale personalization found in algorithmic curation. The article draws on the concept of ubiquitous music and other understandings of the affective and functional aspects of music listening as an everyday practice to reflect upon how Spotify’s approach to datafication of listening potentially leads it to prioritize music recommendations that entice users to engage in inattentive and continuous listening. In extension to this, the article seeks to contribute with knowledge about how the datafication of listening potentially shapes listening practices and conceptions of relevance and quality in music recommendation.

Measuring semantic distances using linked open data and its application on music recommender systems

PurposeMeasuring the similarity between two resources is considered difficult due to a lack of reliable information and a wide variety of available information regarding the resources. Many approaches have been devised to tackle such difficulty. Although content-based approaches, which adopted resource-related data in comparing resources, played a major role in similarity measurement methodology, the lack of semantic insight on the data may leave these approaches imperfect. The purpose of this paper is to incorporate data semantics into the measuring process.Design/methodology/approachThe emerged linked open data (LOD) provide a practical solution to tackle such difficulty. Common methodologies consuming LOD mainly focused on using link attributes that provide some sort of semantic relations between data. In this work, methods for measuring semantic distances between resources using information gathered from LOD were proposed. Such distances were then applied to music recommendation, focusing on the effect of various weight and level settings.FindingsThis work conducted experiments using the MusicBrainz dataset and evaluated the proposed schemes for the plausibility of LOD on music recommendation. The experimental result shows that the proposed methods electively improved classic approaches for both linked data semantic distance (LDSD) and PathSim methods by 47 and 9.7%, respectively.Originality/valueThe main contribution of this work is to develop novel schemes for incorporating knowledge from LOD. Two types of knowledge, namely attribute and path, were derived and incorporated into similarity measurements. Such knowledge may reflect the relationships between resources in a semantic manner since the links in LOD carry much semantic information regarding connecting resources.

Improving the Cold Start Problem in Music Recommender Systems

The music recommender system can help us extract valuable music from a huge amount of raw information, due to the increasingly severe data sparsity and cold start problems in music recommendation system, the recommendation results will be inaccurate. But the traditional algorithm cannot effectively solve these problems; the existing improved algorithm still requires specific parameters in advance due to its poor stability. In this paper, a proposition for a cold start based on community detection algorithm is proposed. By projecting the bipartite network, calculating the similarity between the User and the Item, the Louvain algorithm is used to perform community detection on the projected One-mode network, so that the new record can be updated to the original community, and then the user group is recommended for music. With higher accuracy, solid stability, and shorter running time (when compared with other ease cold start algorithms), the algorithm in the larger community can be safely applied to the music recommendation system or prediction field.