A neural network which generates folk story motifs. Uses Char-RNN-Tensorflow, trained on a cleaned up copy of the Stith Thompson’s Folk Motif Index.  Before training I stripped desired text from the html using beautiful soup.

Some sample output:

Wolf is requested by breaking tabu–miscellaneous.

Enmity between frog and mouse.

Lie: hawk swims country after funeral magic object.

Only woman beauty to get into the besieged cup.

Double descendants of enemy.

Revenant from return of fires.

Lakes of night.

Tree pulls out the sea.

Non-malevolent ghost knowledge.

Eve as an eye.

Trickster, poets alive in heaven.

Enmity between dog and rat.

Why he has built bridge.

Merman lures woman to warm to underground.

Enmity between fowl and water-spirits.

Monk blows horn.

Man helps him (seven) years until he is thrown into water.

Extraordinary book.

Enters church howls.

Today’s catch of food under stone.

Culture hero (demigod) overcomes blood.

Dispute between birds and big ox.

Seduction under water condemned man sacrifice.

Remarkable boar’s foot.

Dragon to open (ashes) of gods.

Approaching and sets fire to much.

Creation of the sun.

Grandfather shows victory and men in a place of holy person.

Dog valley.

Enmity between dog and leopard, elephants.

Ghost walks through city in throat of the world.

Thorn preferred to god.

Person lives on fruit flows from head.

Submarine sound of dead and light.

Truth thought to be ignorant priest.

Queen persuades dupe to suck pool.

Launched: April, 2018

Platforms: Twitter , Mastodon