Today, we usually define a song’s genre by its vocal elemnts – melody, instruments, etc.
We intend to find if we can define a genre simply through the lyrics – whether the written elements of songs contain significant genre information.
Also, this is important to the study of literature and poetry – it could hint at a cognitive connection between melodies and ideas, and help us define the term “genre” and its meanings, thus grounding the study of literature and poetry in something a bit more robust than just personal taste and opinion.
Logistic regression is a type of regression analysis used for predicting the outcome of a categorical dependent variable based on one or more predictor variables.
- Words – we’ll examine how distinct is the vocabulary of each genre.
- Structure – if time constraints will allow, we’ll examine whether genres also imply a certain structure on the lyrics.
- Diversity – even if genres use the same words, do they repeat themselves (“Baby, baby, baby oooh”), or does each song contain a wider selection of words?
Lyrics databases aren’t quality-controlled, don’t carry the same formats between songs (especially obvious with “censored” words, which can be written in different ways – S-it, Sh*t, *%#$, etc.) No existing tree-banks on song lyrics, which limits us to low-level analysis. No efficient data-mining tools – either we create them, or we hand-create all of our data-sets.