Ranking Lyrics for Online Search.

When someone wishes to find the lyrics for a song they typically go online and use a search engine. There are a large number of lyrics available on the internet as the effort required to transcribe and post lyrics is minimal. These lyrics are promptly returned to the user with customary search engine page ranking formula deciding the ordering of these results based on links, views, clicks, etc. However the content, and specifically, the accuracy of the lyrics in question are not analysed or used in any way to determine the rank of the lyrics, despite this being of concern to the searcher. In this work, we show that online lyrics are often inaccurate and the ranking methods used by search engines do not distinguish the more accurate annotations. We present an alternative method for ranking lyrics based purely on the collection of lyrics themselves using the Lyrics Concurrence.