Applying word embeddings to increase usable data

Extracting terms from text is a good starting point for text analysis. With the text tokens we have created so far, we can compare term frequency for different categories, which begins to tell us a story about the content that dominates a particular newsgroup. However, the term alone is just one part of the overall information we can glean from a given term. The previous plot contained people and, of course, we know what this word means, although there are multiple nuanced details connected to this term. For instance, people is a noun. It is similar to terms such as person and human and is also related to a term such as household. All of these details for people could be important but, by just extracting the term, we cannot directly derive these other details. This is where embeddings are especially helpful...