SRBench

SRBench是由萨尔兰大学创建的一个用于评估配乐推荐系统性能的基准数据集。该数据集包含470首来自不同音乐风格的歌曲，旨在通过用户偏好判断来评估推荐系统的性能。数据集的创建过程涉及从Wikipedia和Last.fm等平台收集音乐风格和情感信息，以确保数据集的广泛覆盖性。SRBench的应用领域主要集中在评估和改进配乐推荐系统，旨在解决如何为特定图像集合推荐最合适的背景音乐的问题。

SRBench is a benchmark dataset developed by Saarland University for evaluating the performance of soundtrack recommendation systems. It contains 470 songs spanning various music genres, and is intended to assess the performance of recommendation systems via user preference judgments. The construction of SRBench involved collecting music genre and emotional information from platforms including Wikipedia and Last.fm, so as to ensure the comprehensive coverage of the dataset. The main application areas of SRBench focus on evaluating and improving soundtrack recommendation systems, aiming to solve the problem of recommending the most appropriate background music for a specific image collection.