[This semester I’m taking Humanities Data Analysis with Professor Ben Schmidt. One of our tasks for this week was to build a random-walk generator using 3-grams. Here’s my quick writeup of my generator cross-posted from our course blog.]
We’ve been reading a lot of fairy tales around my house recently, so I wanted to see how well-spun of a tale I could create by walking randomly through a collection of fairy tales. I selected four fairy-tale collections from Project Gutenberg to test this idea on. Code is on GitHub.
I selected these four collections:
- The Thousand and One Nights, Volume 1
- The Blue Fairy Book (by Andrew Lang)
- Fairy Tales of Hans Christian Anderson
- The Fairy Tales of Charles Perrault
The addition of the Arabian Nights stories to Western European fairy tales makes the random generator more interesting, sometimes throwing the geographical sense of the tale into a different place and a different world.
This version generated my favorite beginning: “once upon a time a man by the river yes he was looking straight into the deep waters skeletons of walruses.”
But other versions of the generator took an even darker turn. Here’s the raw text:
“once upon a great procession which was conscious of pain And sore regret of which she said nothing but torment and affliction that He sniffed about to give the ants were always running to and when he approached her they did not really birds but she bore thee Thou hast nothing to me Only tell me something Why this is what you say What is the news O my sister relate to me Art thou she whom he found it impossible to think of The old rough doll You are learned and wise men assembled together in his age and to nail up my mind every earthly care and sorrow with soft turf From the narrow walks and the Wezeer the father of Is both of you should care so much that renders men sinful and impure He fully realized the true the speaker s hand saying to each other till the morning following I have with me from first to last and then burst and fell fast asleep”
And here’s the story, with some punctuation that I added for “clarity”:
Once upon a great procession–which was conscious of pain and sore regret, of which she said nothing but torment and affliction that He sniffed about to give. The ants were always running to, and when he approached her, they did not really birds but she bore thee: “Thou hast nothing to me. Only tell me something: Why this is what you say? What is the news? O my sister relate to me! Art thou she whom he found it impossible to think of? The old rough doll? You are learned, and wise men assembled together in his age and to nail up my mind every earthly care and sorrow with.” Soft turf from the narrow walks and the Wezeer the father of Is, both of you should care so much! That renders men sinful and impure. He fully realized the true the speaker’s hand, saying to each other till the morning following, “I have with me from first to last,” and then burst and fell fast asleep.
And sometimes it’s important to be reminded of where your texts come from. I didn’t remove any text at all from the Project Gutenberg texts, which means that the copyright and distribution information could appear in our stories too. For example:
“The two grand annual festivals are observed with public domain eBooks Redistribution is subject to particular laws or rules with respect to our beetle to himself but the observance of this Wezeer So the porter approached the Distracted Slave of Love when his boat or playing in the lap of prosperity and the fear of him said the Fire drum Peter has gone away I ll do something in me.”
I might publish a longer generated story sometime soon, but all this generator proves is that tales can be wiggly indeed.