Why my true negative is predicted as true positive?

snadibah

i am running a project of sentiment analysis by using SVM. but my result is kind of worrying me. because as i read SVM is expected to be having high accuracy in result. but my result using svm is kind of low. why is this happening? when i added attribute neutral to the sentiment, it became much lower than this. any idea of how to tackle this problem? or is it normal for my true negative result to be predicted as wrong?

MartinLiebig

your model just always says it is positive.