Abstract
We know that text-mining technology is effective to search or to classify documents. Its characteristic consists of using the so-called “co-occurence”; co-occurrence means that two or more similar words appear together. We think that there are many types of information that present this property, music being one them. If we suppose that similar musical notes appear closely, by assigning correspondent words to each note, we believe that text-mining technology can be used to search or classify music. Based on this idea, in the present research we analyzed more than 200 pieces of Japanese popular music. The results obtained by using this analysis were very similar to the opinions that experts on Japanese popular music express intuitively.
