Source of aggressively toned texts for machine learning

Question

I was looking for a source where there would be an extended amount of texts with aggressive tone. It can be intertwined with non-aggressive texts, but it would be good if the ratio of aggressive texts to non-aggressive texts be no less then 1:5.

I was thinking about discussion forums which would be discussing the immigrants issue with contributors of lesser mind who would be very hateful and xenofobic. Or supporters of extreme political parties.

The content being spoken about is not really important, the more diverse it would be the better.

In the past I was looking for a less specific sentiment like "like/dislike" something for which reviews of movies on IMDB were an ideal source and I had some good results.

However it is very difficult to find a base of data with aggressive sentiment or at least some good internet sources from which the data could be gathered.

Do you have any ideas on how I could gather up such data?

I did not know of this one, data science seemed the closest. I will duplicate the question there, but will leave it hanging here for a short while before I delete it, just in case. — Ev0oD, Jul 29 '15 at 13:39
A good source could be banned accounts / deleted messages from boards, though it might be hard to filter spam from it. — Nicolas Raoul, Jul 30 '15 at 10:51

score 1 · Answer 1 · answered Jul 29 '15 at 18:36

1

Depending on how long you need your articles to be, I'd suggest browsing reddit.com for some of the more extreme political or social groups. You should be able to scrape the material fairly easily.

answered Jul 29 '15 at 18:36

Ann Feeney

66
1

2

reddit has an api, so after browsing, you can programmatically download content. Also there is a recent export of nearly all reddit comments. see here – philshem Jul 30 '15 at 01:38

score 1 · Answer 2 · edited Jun 18 '20 at 08:24

Go to Urban Dictionary and look up for any given name
Scroll down to the most down-voted submissions
All of the submissions with five times more downvotes than upvotes have been written to diss a person with that name. Nearly all of them are aggressive.

Examples:

Rebecca is a friendly, loving person at first. But then you will see that she is a manipulating, tarty, lying, backstabbing bitch. She is guaranteed to steal your boyfriend and bullshit about you. She is a stupid slag, she is so far up herself you can hardley see her. Someone needs to tell her where she can shove it, and give her the slap she deserves.

Kevin: The perfect little kitchen boy you will ever meet. Man look at kevin clean that kitchen

It is not open data though: "You may not copy, modify, or reverse engineer any part of the Website"

Source of aggressively toned texts for machine learning

2 Answers2

Linked

Related