1.       Write a computer program that uses the package available at [616] for sentence-level sentiment classification.

2.       Design a framework and architecture for aspect-based sentiment classification with the use of recurrent neural networks.

3.       Design an approach that transforms the problem of text segmentation to that of graph partitioning. Assume that each sentence is a node in the graph. Discuss the various ways in which one might place links between nodes and set their weights. Discuss their advantages and disadvantages. [This is an open-ended question without a uniquely correct answer]