1.       The ranking SVM is a special case of the classical SVM in which each class variable is +1 in the training data (but not necessarily at the time of prediction) and the bias variable is 0. Show that any classical SVM in which the bias variable is 0 but the class variables are drawn from {−1, +1} can be transformed to the case in which each class variable is +1. Why do we need the bias variable to be 0 for this transformation?

2.       Implement an algorithm to discover all 2-skip-2-grams from a given sentence

3.       Consider the following sentence, “The sly fox jumped over the lazy dog.” Enumerate all the 1-skip-2-grams and 2-skip-2-grams.