What does your writing style reveal about you?

Depending on your interests, you can follow one of the three scenarios (or 'types') described below.


The narcissistic type
Have you always wanted to know whether your writing style resembles that of JK Rowling more than that of William Shakespeare? Do you write like a man or like a woman? Are you an introverted type of person or rather an extraverted type?
Paste in a prototypical example of both styles of comparison (for instance, a text sample of Shakespeare and one of JK Rowling) as 'training'. As associated labels, write 'Shakespeare' and 'Rowling', respectively.
Then, paste in a nice example of your own writing as 'test'. Leave the 'label' empty.
Press the button 'classify' below (no need to worry about parameters) and learn the result!

The curious type
You received an anonymous text and want to know which friend wrote it? Do you aim to settle the disputed authorship of the Federalist Papers?
Paste in a text sample of which you know for sure that it is written by, for instance, your friend Alice, and another one by your friend Mad Hatter. Then, paste in the text of which you want to know the authorship, and press 'classify'.

The serious scholar
You are a researcher in Computational Linguistics, Machine Learning, Digital Humanities or another field and want to know how useful automatic analysis of writing style really is.
You can use this demo to perform supervised two-class classification of texts. If you want to process a data set with more classes, send me an e-mail and I can make it happen.


Paste in the texts

Enter your training texts and label (one of each class, max. 2000 words each)

words left associated label:


words left associated label:


Enter the test document of which you want to get the predicted label (max. 2000 words)

words left


Test label

If you know the label for the test document, you can enter it here. It will not be used for classification, but allows for a formal evaluation of the system.

Label for test document:

Parameters (optional)

Task:

Type of features: Show/hide information

Value of n in ngrams: Show/hide information

Number of features in vectors: Show/hide information

Machine Learner: Show/hide information



Kim Luyckx & Walter Daelemans
CLiPS Computational Linguistics Group, University of Antwerp, Belgium

Last update: