<![CDATA[In a song-making, one of the main component which must be considered is lyric. Lyric in a song play a main part to deliver the emotion or meaning from the songwriter to the listener. Sometimes, the emotion means to delivered by the writer is misinterpreted by the listener. To avoid the misinterpreted song-lyric meaning manually, an automatic classification is needed. Classification is also needed to gain information about the emotion from the songs accurately. One of the method used is K-Nearest Neighbor. Before classifications process, there are several steps need to be done such as text pre-processing and weighting using WIDF method. 108 data used in this research with the ratio 1:5; in which, 18 data used for testing and 90 data used for training with the same amount of data each class. The result from 6 attempts of testing based on random K value shown the best average precision is 0,49 and the best recall is 0,53. Songs classification with WIDF weighting method shown a poor accuracy results for 66%. Ambiguity of the words and amount of data training cause the less optimal result.]]>
Copyrights © 2019
