The DAT measures creativity in under 4 minutes. It involves thinking of 10 unrelated words. Creative people choose words with greater semantic distances between them.
See our manuscript in PNAS titled Naming unrelated words predicts creativity for more details.
Try the task and see your score at datcreativity.com.
You can install this library directly from GitHub using pip:
pip install git+https://github.com/jayolson/divergent-association-task.gitDownload and extract the GloVe model file (glove.840B.300d.zip) from https://nlp.stanford.edu/projects/glove/ and place it in your project directory.
Alternatively, you can install manually by following these steps:
- Clone this code:
git clone https://github.com/jayolson/divergent-association-task.git
- Install Python 3 and pip.
- Download the dependencies and model:
make installon Unix-like systems- or else:
pip3 install --user numpy scipy- Download and extract glove.840B.300d.zip from https://nlp.stanford.edu/projects/glove/
- Try it:
python3 examples.py
See examples.py:
import dat
# GloVe model from https://nlp.stanford.edu/projects/glove/
model = dat.Model("glove.840B.300d.txt", "words.txt")
# Compound words are translated into words found in the model
print(model.validate("cul de sac")) # cul-de-sac
# Compute the cosine distance between 2 words (0 to 2)
print(model.distance("cat", "dog")) # 0.1983
print(model.distance("cat", "thimble")) # 0.8787
# Compute the DAT score between 2 words (average cosine distance * 100)
print(model.dat(["cat", "dog"], 2)) # 19.83
print(model.dat(["cat", "thimble"], 2)) # 87.87
# Word examples (Figure 1 in paper)
low = ["arm", "eyes", "feet", "hand", "head", "leg", "body"]
average = ["bag", "bee", "burger", "feast", "office", "shoes", "tree"]
high = ["hippo", "jumper", "machinery", "prickle", "tickets", "tomato", "violin"]
# Compute the DAT score (transformed average cosine distance of first 7 valid words)
print(model.dat(low)) # 50
print(model.dat(average)) # 78
print(model.dat(high)) # 95For open data (8,900 participants from 98 countries), see https://osf.io/kbeq6/.
By Jay Olson at Harvard University.
The dictionary (words.txt) is based on Hunspell by László Németh.
MIT