Mapping American hate speech

This past week we’ve discussed using geotagged Twitter data to create maps, and continued the conversation over using technology to monitor hate speech. Relevant to both of these discussions is a new project from geographers at Humboldt State university to create Twitter heat maps showing showing the use of slurs against racial minorities, gays, and people with disabilities in the United States.

As I discussed in the original Hatebase post, monitoring hate speech can be difficult since the same word can often have different connotations depending on its context and, sometimes, who’s using it. The Humboldt State researchers were criticized for not taking this into account in a previous project to map racist Tweets directed at Barack Obama following his election. 

This time, they were a bit more careful, which meant some poor students had the unenviable task of reading though thousands of Tweets to see which one’s were actually derogatory. Professor Monica Stephens writes:

Using DOLLY to search for all geotagged tweets in North America between June 2012 and April 2013, we discovered 41,306 tweets containing the word ‘nigger’, 95,123 referenced ‘homo’, among other terms. In order to address one of the earlier criticisms of our map of racism directed at Obama, students at Humboldt State manually read and coded the sentiment of each tweet to determine if the given word was used in a positive, negative or neutral manner. This allowed us to avoid using any algorithmic sentiment analysis or natural language processing, as many algorithms would have simply classified a tweet as ‘negative’ when the word was used in a neutral or positive way. For example the phrase ‘dyke’, while often negative when referring to an individual person, was also used in positive ways (e.g. “dykes on bikes #SFPride”). The students were able to discern which were negative, neutral, or positive. Only those tweets used in an explicitly negative way are included in the map.

All together, the students determined over 150,000 geotagged tweets with a hateful slur to be negative.

The tweets are then normalized based on the total amount of Twitter activity in the surrounding area. They note that while “Orange County, California has the highest absolute number of tweets mentioning many of the slurs,” it doesn’t show up all that prominently on any of the maps because of its high overall volume of tweets. The map above shows all the racist terms tracked.

Even after normalization, most of the maps just look an awful lot like U.S. population density maps.  It’s hard to draw any conclusions from these on where animosity against a particular groups is most prevalent.

Depressingly, there is one notable exception:

Via the Guardian‘s Data Blog.