Wednesday, October 29, 2014

Sentiment Mapping


Twitter is a great resource for researching language use and people's moods. The Geography of Hate, by Dr. Monica Stephens of Humboldt State University, is one of the best examples of a map which analyses Twitter messages to identify specific sentiments.

The Geography of Hate map shows the rough location of every geocoded tweet in the United States, from June 2012 - April 2013, which contained one or more of ten 'hate words'. Users of the map can view three different heat maps, one for homophobic tweets, one for racist tweets and one for anti-diasablity tweets. The user can also view individual heat maps for any one of the ten offensive words.


London Feels is a map visualizing how Londoners feel based on their Twitter messages. The map shows the location of Twitter messages inside the M25 which contain a number of key words indicating some kind of sentiment (e.g. 'terrible', 'bad', 'good', 'awesome' etc).

Positive tweets are shown on the map in blue, negative tweets shown in red and average feeling tweets are shown in purple. Unlike The Geography of Hate map there is no human analysis of the Twitter messages. Therefore the map obviously shows a lot of messages where people are expressing a judgement upon something rather than a sentiment about their own state of being (e.g. 'that movie is terrible').


A more scientific approach is taken by Mappiness. Two academics at the London School of Economics are carrying out research into how the environment affects people's happiness. To help them gather the data for this research they launched an iPhone app to track how people are feeling.

If you download the app you will receive a notification between one and five times a day. The notification will ask you to open the app, briefly report how you're feeling and who you're with, where you are, and what you're doing. If you're outdoors and you're happy, you'll also be asked to take a photo of your surroundings.

Anyone can keep an eye on the research being gathered by checking out the Mappiness Google Map. The map shows the outdoor places where Mappiness users have most recently reported feeling happy. If you click on an information window you can also view the photograph taken at that happy moment.


Weather Sentiment Prediction is a clever Google Maps based application that can tell you how people feel about the weather at any location.

You can select a location on the application by dragging the map marker and adjusting the radius of the search area. Once you hit the 'search' button the application begins to analyse Tweets from that area which mention the weather. Each message is machine analysed to determine whether it is a positive or negative response to the weather.

A smiley face and an unsmiley face at the top of the Twitter stream give an indication of the overall response to the current weather.

No comments: