Sentiment Analysis is a branch of research from Text Mining which is used to claccify text document in the from of opinion based on sentiment. The research about Sentiment Analysis has been done with using several data mining method. One of them is doing classification Sentiment Analysis with using Support Vetor Machine method based on Particle Swarm Optimization. The kind of text that is used in the research is comments on the result of lecturer evaluations on SIMARI. The best accuracy performance use Support Vector Machine algorithm with Particle Swarm Optimizationis gained by classification examination with parameter values cost worth 1 with result 82,26%, then on examination with Particle Swarm Optimization use particle amount 30 and 10 iteration, are obtained with the highest accuracy result of 82,59%.