Stop N-gram Removal and Topic-Similarity to Improve Topic Modeling

In order to enhance the performance of topic modeling algorithms and determine the optimal number of topics, this thesis will present two different approaches: Stop N-gram Removal, a novel preprocessing procedure based on the elimination of a dynamic number of repeated words in text documents and Topic-Similarity, a new way to determine the optimal number of topics which automatically measures the similarity between the meanings of the words in each topic.