Voyant recently gained a new visualization tool; the Constellation Viewer.
The Constellation viewer is a tool designed to let users explore Word2Vec models. A Word2Vec model is any mathematical model that lets users map words (or tokens) into vector space, and is a fundamental technology that is used by all modern language based AI technology. Essentially, words that have similar vectors also have similar meaning.
One limitation of Word2Vec visualizations is that these models generally map words into high dimensional vector space (100 or more dimensions) while we can only visualize data using two. What the constellation viewer does is calculate the distance between words in a corpus and uses a network graph to connect words that the model views as similar to each other. This way a user can explore similar meaning in words.
Currently, we can convert tokens to vectors using basic analysis techniques such as PCA, correspondence analysis, and t-SNE. However, in the future we hope to add support for more robust word2vec algorithms.
See the demo implementation here: https://voyant-tools.org/?corpus=austen&view=Constellations
See the documentation here: https://voyant-tools.org/docs/tutorial-constellations.html