Natural language processing (NLP) techniques can learn the key semantic elements contained in large collections of text in a completely unsupervised manner allowing for the automatic discovery of trending topics, anomalous event or hidden correlations. In this video, T. J. Hazen discusses how NLP techniques for learning key phrases and latent topics can be applied to text corpora, and demonstrates how these techniques perform on two different data sources: (1) newswire data and (2) the US Congressional Record.