Machine Learning & Data Science Natural Language Processing One hot Word Representation

One way to represent a word in your vocabulary as a vector is to use one-hot vectors. One-hot vectors are binary vectors where all elements are zero except for one element, set to one, which corresponds to a particular word.

Pros:

• Simple
• No implied ordering Cons:
• Huge vectors, scaling with vocabulary size
• No embedded meaning; can't compare words meaningfully.