音乐推荐系统主观评价指标研究——以网易云音乐为例

Study on the Subjective Evaluation Metrics System of Music Recommendation System: A Case Study of NetEase Cloud Music

算法推荐在大数据时代是用户获取信息的重要渠道,推荐质量关乎用户的体验态度。如今在用户基数庞大的音乐推荐系统背后,推荐质量参差不齐、用户心理感受被忽视,该领域缺乏一个通用的用户评价指标体系。因此,研究以用户心理体验为导向,在文献回顾和深度访谈的基础上,编制出适用于音乐推荐系统评价的初始量表,通过预调查和正式调查,并采用探索性因素分析、验证性因素分析等统计方法对调查结果进行检验分析,最终获取由内容性、功能性、页面设计、感知易用性、感知体验、干扰风险和质量风险七个维度构成的音乐推荐系统用户评价量表。研究进一步对网易云音乐用户的推荐系统体验态度以及相关影响因素进行探索,最后从实践角度对音乐平台推荐系统的优化方向提出建议。

The algorithm-based recommendation provides an important access for users to obtain information in the era of big data.Recommendation quality affects users’ attitude towards experience.Given the large user base that music recommendation system are applying to,there exist problems such as uneven quality of music recommendation,ignorance of users’ feelings,and the absence of a universal evaluation metrics system.This study focuses on the users’ psychological experience,based on the literature review and in-depth interviews,the initial scale for the evaluation of music recommendation system is proposed.Statistical analysis methods such as exploratory factor analysis and confirmatory factor analysis are adopted to test the results of pre-survey and formal research.We develops an evaluation scale suitable for the music recommendation system,including seven dimensions,i.e.content,functionality,page layout,perceived ease of use(PEOU),perceived experience,interference risk and quality risk.This study further explores the NetEase cloud music users’ attitudes towards recommendation system and influence factors.Finally,from the practical point of view,suggestions for optimization of the music recommendation system are discussed.