CAME: Content- and Context-Aware Music Embedding for Recommendation

Abstract

Traditional recommendation methods suffer from limited performance, which can be addressed by incorporating abundant auxiliary/side information. This article focuses on a personalized music recommender system that incorporates rich content and context data in a unified and adaptive way to address the abovementioned problems. The content information includes music textual content, such as metadata, tags, and lyrics, and the context data incorporate users' behaviors, including music listening records, music playing sequences, and sessions. Specifically, a heterogeneous information network (HIN) is first presented to incorporate different kinds of content and context data. Then, a novel method called content-and context-aware music embedding (CAME) is proposed to obtain the low-dimension dense real-valued feature representations (embeddings) of music pieces from HIN. Especially, one music piece generally highlights different aspects when interacting with various neighbors, and it should have different representations separately. CAME seamlessly combines deep learning techniques, including convolutional neural networks and attention mechanisms, with the embedding model to capture the intrinsic features of music pieces as well as their dynamic relevance and interactions adaptively. Finally, we further infer users' general musical preferences as well as their contextual preferences for music and propose a content-and context-aware music recommendation method. Comprehensive experiments as well as quantitative and qualitative evaluations have been performed on real-world music data sets, and the results show that the proposed recommendation approach outperforms stateof-the-art baselines and is able to handle sparse data effectively.

Sequential Recommendation Based on Multivariate Hawkes Process Embedding With Attention

“…An attention [41], [42] is intuited from visual attentions of human beings (incline to be attracted by more important parts of a target object). Attention is widely used in many fields, including object detection [43], [44], prediction [45], query suggestion [46], and recommendation [4]. In brief, attention can be used to increase the interpretability and adaptivity of complex models such as neural networks by calculating the weights of different data/information automatically.…”

“…Xiao et al [49] propose a model named Attentional Factorization Machines, which uses an attention model and Factorization Machines to model the importance of different features and their interactions. Wang et al [4] present a content-and context-aware recommendation model called CAME, which use an attention model and Convolutional Neural Network to learn the content features adaptively for music recommendation. Especially, an attention mechanism enables CAME to model different aspects of music and enhanced its ability of capturing music pieces' dynamic features.…”

“…Generally, traditional recommendation methods include collaborative filtering, content-based, context-aware and hybrid models [3]. All kinds of recommendation methods are widely applied in many fields, such as music recommendation [4], point of interests (POI) recommendation [5], [6], groups recommendation [7], [8], business process recommendation [9], E-Commerce [10] and so on. Most traditional recommendation techniques perform recommendation based on users' general interest, which represents users' long-term static preferences.…”

DFIAM: deep factorization integrated attention mechanism for smart TV recommendation

“…At present, recommender system is developed to solve this problem. It is widely used in video [14,30], social media [17,31], music [24], and e-commerce [13,16,25], etc. And it utilizes users' history behavior to improve the effectiveness of searching and to retrieve interesting items.…”

Time-aware sequence model for next-item recommendation

“…Consequently, it becomes increasingly difficult to find the information they really need. Therefore, recommender systems have been proposed to help users find the contents that they want, such as research articles [8], Point-of-Interest [20,32], question [4] and music [23,24]. Existing recommendation methods include collaborative filtering-based recommendations [12,29], content-based recommendations [16], social network-based recommendations [13,1], and hybrid recommendation [7].…”