The idea of this project is to create a poem based on dreams I have had. Using natural language processing through Spacy Library the program extract the nouns, verbs and adjectives and mix them in a different sintaxis.

My interest it to explore the meaning/interpretation that arise from this re-construction of words and speech.

I´ve been writing my dreams in notebooks since 1995 and on 2016, when I move to live to NYC i decided to digitalize them. At ITP I started a researching process experimenting with dreams in different computation media and programs.

To create this work I made a json file with an array of dreams I had during 2016.

In the program I import Spacy to work with Natural Language Programming (nlp). I created set() of verbs, nouns and adjectives that collects the verbs, nouns and adjectives of all the dreams that are in the json file. Also list() of verbs, nouns and adjectives that allows the program to extract the verbs, nouns and adjectives of single dreams from the array.

First I started exploring mixing all the dream´s words to create new strophes, but I felt that through that process the meaning became too wide.

 


 

 

 

 

Then I decided to use one dream for each strophe to keep a sense of the stories on them.

Then I started playing with the shapes. In a previous homework I created a module with some functions to do this.

#create empty lines between sentences

def multi_lines(n):
newString=””
for i in range(n):
newString +=”\n”
return newString

#create spaces between words
def multi_spaces(n):
newString=””
for i in range(n):
newString +=” ”
return newString

I realice that on 2016 I had a lot of dreams about crossing. I guess it was because I was moving from one city/country to another. That´s why the poem is called Cross, and I tried to create a shape that expreses that movement.

 

I explored the idea of adding the words “I am” before each noun, verb or adjective, based on the popular idea that in dreams the dreamer is each character.

 

I test it, but I didn´t like the result. So I came back to my original idea.

 

The program: