Administered two models to solve the task - One based on Dependency Trees to exploit underlying sentence structure thus eliminating the need for any training data & the other one is based on LSTM networks with attention-mechanism. The models have been trained & evaluated on abstract-title pairs collected from arXiv.org