Message Boards Message Boards

[WSS18] Authorship Identification Using Neural Networks

POSTED BY: Jaime Buitrago
9 Replies
Posted 6 years ago

This is really nice, Jaime. I am trying to reproduce your results. Could you provide me the two files plotSet.mx and plotSetClas.mx? My email is ruifeng14@gmail.com

Thank you!

POSTED BY: Ruifeng Dong

Good to know, that makes sense. Looks like someone was attempting it already: https://blog.bookmatch.com/analysis/2018/09/09/who-wrote-the-resistance-op-ed.html

POSTED BY: Gavin Holt

Definitely, it can be used. However, the challenge is to find training data, e.g. writings from all the "suspect" authors. In any case, we would just get a "writing style" verification, but that would be interesting any way.

POSTED BY: Jaime Buitrago

Very nice! Could it be set-up to attempt to identify the anonymous author of the NYT piece about the White House?? :-p

POSTED BY: Gavin Holt

Yes, definitely. You would need writings by your suspect authors so that you can train the neural network on the style of each author. But once it is trained, it should be able to determine who among the suspects the culprit is.

POSTED BY: Jaime Buitrago

There are op eds out there by most of the people who were named as potential authors, as well as by the actual author. So you could use those for training data.

POSTED BY: Daniel Lichtblau

enter image description here - Congratulations! This post is now a Staff Pick as distinguished by a badge on your profile! Thank you, keep it coming!

POSTED BY: Moderation Team

This is really nice. And getting recognition accuracy of 66-67% on CCAT-50 is no easy matter.

POSTED BY: Daniel Lichtblau

Thank you very much. I had a lot of guidance from my mentor at Wolfram Summer School @Sebastian Bodenstein

POSTED BY: Jaime Buitrago
Reply to this discussion
Community posts can be styled and formatted using the Markdown syntax.
Reply Preview
Attachments
Remove
or Discard

Group Abstract Group Abstract