Introduction–From Genres to Tags: A Little Epistemology of Music Information Retrieval Research

Introduction–From Genres to Tags: A Little Epistemology of Music Information Retrieval Research
复制标题

引言——从流派到标签:音乐信息检索研究认识论一点

DOI:
--
复制
发表时间:
2008
期刊:
影响因子:
--
通讯作者:
E. Pampalk
E. Pampalk
中科院分区:
--
文献类型:
--
作者:
J. Aucouturier;E. Pampalk

文献摘要

被引文献

相似文献

惊喜:这并没有像预期的那样发生。大多数信息都是手动注释的(没有自动分析),非结构化的(没有分类法),在一个协作的,动态的和无节制的过程中(不像集中式图书馆)。数以百万计的用户经常连接到网站,如www.example.com,MusicStrands,MusicBrainz或Pandora,在那里他们输入他们喜欢或不喜欢的音乐的免费描述(又名标签)。每个用户的标签都可供所有人查看,并影响其他用户描述或查找音乐的方式。其结果是一个合作的音乐知识库的规模和丰富性闻所未闻。“披头士”曾经是“英国流行”。它们现在类似于图1。
Surprise: This didn’t quite happen as expected. Most of the information is annotated manually (no automated analysis), unstructured (no taxonomy), in a collaborative, dynamical and unmoderated process (unlike a centralized library). Millions of users routinely connect to web-sites such as last.fm, MusicStrands, MusicBrainz or Pandora, where they enter free descriptions (aka tags) of the music they like or dislike. Each user’s tags are available for all to see and influence the way other users describe or look for music. The result is a collaborative repository of musical knowledge of a size and richness unheard of so far. ‘‘The Beatles’’ used to be ‘‘British pop’’. What they are now is something akin to Figure 1.