Art&Emotions Dataset

Art&Emotion experiment description  

The Art & Emotions dataset was collected in the scope of EU funded research project SPICE (https://cordis.europa.eu/project/id/870811) with the goal of investigating the relationship between art and emotions and collecting written data (User Generated Content) in the domain of arts in all the languages of the SPICE project (fi, en, es, he, it). The data was collected through a set of Google Forms (one for each language) and it was used in the project (along the other datasets collected by museums in the different project use cases) in order to train and test Emotion Detection Models within the project.
 
The experiment consists of 12 artworks, chosen from a group of artworks provided by the GAM Museum of Turin (https://www.gamtorino.it/) one of the project partners. Each artwork is presented in a different section of the form; for each of the artworks, the user is asked to answer 5 open questions: 

1. What do you see in this picture? Write what strikes you most in this image. 

2. What does this artwork make you think about? Write the thoughts and memories that the picture evokes.

3. How does this painting make you feel? Write the feelings and emotions that the picture evokes in you 

4. What title would you give to this artwork?

5. Now choose one or more emoji to associate with your feelings looking at this artwork. You can also select "other" and insert other emojis by copying them from this link: https://emojipedia.org/

  
For each of the artworks, the user can decide whether to skip to the next artwork, if he does not like the one in front of him or go back to the previous artworks and modify the answers. It is not mandatory to fill all the questions for a given artwork.

The question about emotions is left open so as not to force the person to choose emotions from a list of tags which are the tags of a model (e.g. Plutchik), but leaving him free to express the different shades of emotions that can be felt.  

Before getting to the heart of the experiment, with the artworks sections, the user is asked to leave some personal information (anonymously), to help us getting an idea of the type of users who participated in the experiment. 

The questions are: 

1. Age (open) 
2. Gender (male, female, prefer not to say, other (open)) 
3. How would you define your relationship with art? 

• My job is related to the art world 
• I am passionate about the art 
• I am a little interested in art 
• I am not interested in art 

       4. Do you like going to museums or art exhibitions? 

• I like to visit museums frequently 
• I go occasionally to museums or art exhibitions 
• I rarely visit museums or art exhibitions 

---------------------

Dataset structure:

• FI.csv: form data (personal data + open questions) in Finnish (UTF-8)
• EN.csv: form data (personal data + open questions) in English (UTF-8)
• ES.csv: form data (personal data + open questions) in Spanish (UTF-8)
• HE.csv: form data (personal data + open questions) in Hebrew (UTF-8)
• IT.csv: form data (personal data + open questions) in Italian (UTF-8)
• artworks.csv: the list of artworks including title, author, picture name (the pictures can be found in pictures.zip) and the mapping between the columns in the form data and the questions about that artwork 
• pictures.zip: the jpeg of the artworks