#More Text Analysis with ALICE
Dive deeper into the text analysis looking glass with Alice!
Instructions:
Take the Alice in Wonderland exercise and start working on the two challenge problems there. Once you have a system for analyzing the properties of book-length texts and a system for generating random text based on those properties, do the following:
- Modify your text analysis code to distinguish between text written by humans and text generated randomly by your book-generator program. You can use the public-domain works from Project Gutenberg as examples of genuine human writing (note that "Ulysses" by James Joyce may not be the best choice of training data).
- Once you have a program capable of determining that, say, "Huck Finn" was written by a human and "Red Adventure Sunshine Hat" was generated randomly by a machine, improve your random text generator so that it fools your detector.
- Pit the detector and the generator against one another in a kind of Turing Test arms race, making successive improvements to each so that one side gains the upper hand, and then the other.
- How does your detector handle text in languages other than English? If you used Project Gutenberg as your training data, how well does your detector handle more modern texts like blog posts or tweets?
- Feed the detector one of your term papers from your senior year; are you human? How do you feel about the result?
- Read about Searle's Chinese Box.
View More Text Analysis with ALICE on Learn.co and start learning to code for free.