Java SVM (using weka API or LIBSVM)

I need a program in Java that uses Support vector machines as a meta-learner. The input information will be a database containing 100 given datasets with 5 meta-features for each dataset. The program will run 8 clustering algorithms on each dataset and rank those algorithms for each dataset based on the accuracy, latency and reliability. Afterwards, a new unfamiliar dataset is used and based on the similarity between the meta-features of the new dataset and the meta-features in the database, the program is supposed to predict a ranking for the 8 clustering algorithms for that new dataset.