A Tamil Lyrics Search and Visualization System

The availability of Tamil lyrics available on the web is growing rapidly. It has been difficult for the user to search for lyric characteristics, such as emotion, genre, rhyming features, pleasantness and rhetorical style, from a large data set. In order to increase the interestingness of Tamil lyrics to the user, we have developed lyric visualization using statistical modeling. A collection of 10000 lyrics from various Tamil lyrics books and web sites is used in order to explore the internal and external features of the individual lyric. These data have been incorporated into a multi-search engine, in which, the user will have the facility to use keywords search based on a well-known “weighting” factor called term frequency–inverse document frequency (TF-IDF), a semantic search based on the Universal Networking Language (UNL) and a lyricist search based on features style model. However, it is a challenging task to visualize each lyric with different characteristics. So, we have designed a flower like structure, which has been created using a graphics2D tool, to visualize each lyric characteristic and increase the interest of the user in the lyric search. Moreover, different statistical approaches have also been handled to identify the lyric characteristics. The proposed visualization has been evaluated using various parameters to distinguish the comfort and inclination of the user. The relevance of the document displayed in the search, and the accuracy of the genre and emotion have been characterized, using a precision and recall mechanism.