CS224N Final project: Language Model Methods and Metrics

We first present skip n-grams interpolated with various other n-grams and measure their ability to improve performance of the language models in terms of perplexity. We then present a language model that relies on content words, words which are uncommon. We then present a bag generation algorithm, and use bag generation as a metric for our language models. We then present a language model that uses clustering by part of speech tags.