Z Geography has a great post about the positives and critical flaws of using social media to map things like happiness, much like the Vermont Complex Systems Center did for the Geography of Happiness According to 10 Million Twitter Tweets.
Most of the flaws, but not all, are tied to using a bot scripted language to determine if a tweet is a happy or sad tweet. While this sounds simple enough how would something like "I love you to death" score? It is a very happy tweet but death is usually scored as a negative. Meanwhile how does a bot deal with sarcasim? "I just had the best xp of my life, in traffic for two hours", is a strong negative emotion but would probably count as positive. Plus things like weather can impact what a person tweets but hardly impacts a persons over all happiness level.
Right away Z points out that too many people are taking this simple GIS project and claiming it is finished geography. GIS is a great tool, it is not a finished product though. Geography, and common sense with some critical thinking, is needed to properly analyze GIS output.
Most of the flaws, but not all, are tied to using a bot scripted language to determine if a tweet is a happy or sad tweet. While this sounds simple enough how would something like "I love you to death" score? It is a very happy tweet but death is usually scored as a negative. Meanwhile how does a bot deal with sarcasim? "I just had the best xp of my life, in traffic for two hours", is a strong negative emotion but would probably count as positive. Plus things like weather can impact what a person tweets but hardly impacts a persons over all happiness level.
Right away Z points out that too many people are taking this simple GIS project and claiming it is finished geography. GIS is a great tool, it is not a finished product though. Geography, and common sense with some critical thinking, is needed to properly analyze GIS output.
No comments:
Post a Comment