For this week’s assignment, I chose to create generative text project. To find inspiration for it, I browsed through some data on Kaggle, eventually landing on a dataset of Tweets, where each tweet corresponds to a certain emotion(https://www.kaggle.com/datasets/nelgiriyewithana/emotions). Hence, I tried to use the data to create a somewhat meaningful conversation between two people.
First I uploaded an image of two characters speaking and thought about putting text over their heads like comic book characters, to indicate speech. The speech , in this case, is simply one instance of the Tweets in the dataset, but every Tweet is written from first perspective, giving an illusion of fluency in the conversation. The main idea behind creating a conversation was to randomly choose what the first person has to say, then identify the emotion the speech is tied to, and finally, get another random Tweet from the database that has the same emotion. The only addition to the text would be the connectives words I added manually, but they are also randomly taken from the given array that I wrote.
let openers = ["True, " , "Right!, " , "Agreed, " , "Same, ", "Fax, " , "Relatable, " , "Yeah, " , "I see, "]; //additions to the second speech
I made it easier for me to correlate text with emotions by creating arrays of text for each emotion, so that the random function only has to choose between texts of the same emotion. Example of a conversation below. With each run of the program, different texts and emotions will be generated.
In the future I would like to try to make the conversation make more sense by adding additional variable and edits into the texts, but I feel like sometimes the generated outcome with this code is funny enough to be worth it.