Sculpting Emotions in Code: Generating and Evaluating Micro-Fiction
Submission Type
Paper
Start Date/Time (EDT)
18-7-2024 2:15 PM
End Date/Time (EDT)
18-7-2024 3:15 PM
Location
Algorithms & Imaginaries
Abstract
Flash Fiction, a unique form of literature frequently used in creative writing, challenges writers to craft complete stories within specific constraints, such as a certain letter, number of words or sentence count. With its widespread availability online, this genre provided an ideal opportunity to build a database for a generative system. The brevity of these narratives is closely intertwined with the emotional response readers experience, which is subjected to their interpretation and appreciation. This paper presents a concise overview of the Flash Writer – a system developed in Processing and the RiTa Library, designed to generate extremely short stories under the constraints of a two-sentence story with no more than 280 characters (also known as Twitterature). The paper also explores the system’s creativity by assessing its output through a survey that evaluated the reader’s perception of creativity, language, structure, and chiefly, affective computing.
In 2021, a small online survey was conducted with 21 participants of diverse backgrounds. The survey supplied valuable feedback on the system’s output, as well as valuable input for its enhancement despite the small sample size. Furthermore, the experiment evaluated the reader’s cognitive and emotional responses to a series of micro-narratives created by Twitter (now X) users and the Flash Writer. The stories used in the experiment were curated by the researcher who selected flash stories online and generated others with The Flash Writer. Readers were asked to guess if the stories were generated by a computer system or created by a human agent, as well as give their appreciation on each story. This preliminary research aims to use the reader’s feedback to improve the system further and explore its integration with GPT-4.
Key-words: Generative Literature; Computational Creativity; Affective Computing;
Recommended Citation
Chinellato Ferreira, Mariana, "Sculpting Emotions in Code: Generating and Evaluating Micro-Fiction" (2024). ELO (un)linked 2024. 6.
https://stars.library.ucf.edu/elo2024/algorithmsandimaginaries/schedule/6
Sculpting Emotions in Code: Generating and Evaluating Micro-Fiction
Algorithms & Imaginaries
Flash Fiction, a unique form of literature frequently used in creative writing, challenges writers to craft complete stories within specific constraints, such as a certain letter, number of words or sentence count. With its widespread availability online, this genre provided an ideal opportunity to build a database for a generative system. The brevity of these narratives is closely intertwined with the emotional response readers experience, which is subjected to their interpretation and appreciation. This paper presents a concise overview of the Flash Writer – a system developed in Processing and the RiTa Library, designed to generate extremely short stories under the constraints of a two-sentence story with no more than 280 characters (also known as Twitterature). The paper also explores the system’s creativity by assessing its output through a survey that evaluated the reader’s perception of creativity, language, structure, and chiefly, affective computing.
In 2021, a small online survey was conducted with 21 participants of diverse backgrounds. The survey supplied valuable feedback on the system’s output, as well as valuable input for its enhancement despite the small sample size. Furthermore, the experiment evaluated the reader’s cognitive and emotional responses to a series of micro-narratives created by Twitter (now X) users and the Flash Writer. The stories used in the experiment were curated by the researcher who selected flash stories online and generated others with The Flash Writer. Readers were asked to guess if the stories were generated by a computer system or created by a human agent, as well as give their appreciation on each story. This preliminary research aims to use the reader’s feedback to improve the system further and explore its integration with GPT-4.
Key-words: Generative Literature; Computational Creativity; Affective Computing;
Bio
Mariana Chinellato Ferreira is a seasoned professional with expertise in Languages and Literature. In addition, she brings professional academic experience, both in teaching and researching, in the fields of International Literature, Theory of Literature, Digital Literature, Digital Humanities, Academic and Creative Writing, and Communication.
She holds a dual undergraduate degree in Social Communication and Modern Languages and Literature, which she earned from the Catholic University of Santos – UniSantos. Furthermore, she has achieved an MSc in Sciences from the University of São Paulo – São Carlos.
She is pursuing a doctoral degree from the University of Coimbra, in the Programme of Materialities of Literature. Her research is particularly interested in Creative Generative Systems and she is engaged in collaborative research with the Centre for Informatics and Systems of the University of Coimbra. Her academic journey reflects her dedication to advancing the boundaries of knowledge in her chosen fields, mainly fostering unusual connections.
Currently, she resides in the Netherlands, where she is a Lecturer in English Communication, Academic Writing, and AI. Additionally, she provides coaching and mentorship to first-year international students within the Academy of Built Environment and Logistics at Breda University of Applied Sciences – BUas.