LyricSIM

LyricSIM是一个专为西班牙歌曲歌词语义相似性检测设计的新数据集，由萨拉曼卡大学等机构创建。该数据集最初包含2775对西班牙歌曲，经过63名母语为西班牙语的标注者的集体标注实验，最终精炼得到676对高质量标注数据。数据集涵盖了主题、信息、情感、字面意义和文化背景等多个维度的相似性评分。LyricSIM旨在推动歌曲推荐、搜索和文化分析等领域的研究，特别是在理解和建模西班牙语世界中音乐和歌词的语义相似性方面具有重要意义。

LyricSIM is a novel dataset specifically developed for semantic similarity detection of Spanish song lyrics, created by institutions such as the University of Salamanca. Initially comprising 2,775 Spanish song lyric pairs, the dataset was refined into 676 high-quality annotated pairs via collective annotation experiments carried out by 63 native Spanish speakers. It includes similarity scores across multiple dimensions including theme, information, emotion, literal meaning, and cultural background. LyricSIM is intended to promote research in fields like music recommendation, search, and cultural analysis, and holds great significance for understanding and modeling semantic similarity of music and lyrics in the Spanish-speaking world.