This repository shows four programming assignments done for natural language processing.
The first one is building a modern phrase-based statistical machine translation system. It has two main components: implementation of the IBM word alignment models and feature engineering for an MT decoder.
The second one implements a parsing algorithm for a broad coverage statistical treebank parser and test the algorihtm on the WSJ section of the Penn Treebank.
The third one implements two coreference resolution systems: the first system is rule-based while the second one is based on a discriminative statistical classifier.
The fourth one implement a neural network for named entity recognition: including the word embedding layer, the feedforward neural network and the corresponding backpropagation training algorithm.