We saw the incredible power of deep learning first hand – we could successfully train a neural network to generate text that very much resembles human-produced text, if at least in its syntax and to some extent, grammar and spelling. With more fine-tuning and maybe a little bit of human supervision, we can see how we can create very realistic chatbots with this kind of technology.
While this kind of text analysis may not seem particularly useful for us, neural networks find a lot of use in more practical text analysis tasks, such as in text classification or text clustering. We will be exploring these kinds of tasks in our next chapter – in particular, text classification using Keras and using spaCy.
We present the following links to the reader before moving on to the next chapter; they are blog posts discussing effective strategies when dealing with text...
