I love listening to music and was really curious to know how do musicians evolve emotionally with time and number of records. I selected a couple of artists and downloaded the lyrics for all their published albums. I used the song titles in an album and gave a sentiment score to the words. The words were ranked on the scale of [-5, +5], -5 being the far end of negative emotion and +5 being the other end.
I added color to the words to signify the emotion that goes clockwise from Positive (Blue) to Negative (Red). In the third quarter of the circle the albums for the artist is drawn as smaller circles. Each album is then linked to the words contained in the song titles that are a part of that album. On hover over these circles, Name of the album, Track list and Release Year is displayed on the top left.
My initial idea was to use song lyrics instead of song titles for every album and visualize the emotion using words in the lyrics, more or less like sentiment analysis on paragraphs. I drew some initial sketches to see how would the visualization look like if I managed to get all the data processed and scored for sentiment.
The data had to be cleaned, to remove the stop words in English and break the words in root form. I wrote a code in Java to clean the code and store the results in the MySQL database.