[OC] WebMood.ai
The Internet’s Emotional Barometer
September 27, 2024
[OC] WebMood.ai The Internet’s Emotional Barometer
Posted by icyou520
5 comments
What an interesting concept and website. Bookmarked so I can see how it goes crazy on and after Election Day!
I find it interesting that all those sources are considered “news” rather than just propaganda for thier various political sides.
A correlation between negativity/ positivity and bias might be more useful because that isn’t immediately obvious.
It appears there may be some misunderstandings about what this data represents, so I’d like to clarify. I’m not influenced by politics in this analysis.
I collect the top headlines from each website—currently 158 and growing—and perform a sentiment analysis on them. Each headline is labeled as Positive, Neutral, or Negative, and assigned a corresponding score.
For example, if Breitbart has 10 news stories—6 about wars, 2 about healthcare, and 2 about election fraud—the war-related headlines are likely classified as Negative, resulting in a higher negative score for the site.
In contrast, if CNN has 10 stories—2 about war, 6 about fashion, and 2 about puppies—the positive topics like fashion and puppies lead to a higher positive rating.
You might argue that Breitbart covers more challenging stories without positive ones to balance them out. However, all websites fluctuate between negative and positive scores over time; even CNN had a negative score a couple of days ago.
The chart simply reflects the emotion or tone of each website at the current moment. I’m interested in observing over the course of a year which sites tend to be more positive or negative overall.
Why is it sorted largest to smallest twice? It either needs a gap or sorted consistently
How is sentiment analysis performed?
I’m wondering if sentiment is defined in such a way that one would expect the average sentiment of a corpus to be neutral. In other words, does this result fall out of the definition? How do other corpuses stack up when this analysis is performed?
5 comments
What an interesting concept and website. Bookmarked so I can see how it goes crazy on and after Election Day!
I find it interesting that all those sources are considered “news” rather than just propaganda for thier various political sides.
A correlation between negativity/ positivity and bias might be more useful because that isn’t immediately obvious.
It appears there may be some misunderstandings about what this data represents, so I’d like to clarify. I’m not influenced by politics in this analysis.
I collect the top headlines from each website—currently 158 and growing—and perform a sentiment analysis on them. Each headline is labeled as Positive, Neutral, or Negative, and assigned a corresponding score.
For example, if Breitbart has 10 news stories—6 about wars, 2 about healthcare, and 2 about election fraud—the war-related headlines are likely classified as Negative, resulting in a higher negative score for the site.
In contrast, if CNN has 10 stories—2 about war, 6 about fashion, and 2 about puppies—the positive topics like fashion and puppies lead to a higher positive rating.
You might argue that Breitbart covers more challenging stories without positive ones to balance them out. However, all websites fluctuate between negative and positive scores over time; even CNN had a negative score a couple of days ago.
The chart simply reflects the emotion or tone of each website at the current moment. I’m interested in observing over the course of a year which sites tend to be more positive or negative overall.
Why is it sorted largest to smallest twice? It either needs a gap or sorted consistently
How is sentiment analysis performed?
I’m wondering if sentiment is defined in such a way that one would expect the average sentiment of a corpus to be neutral. In other words, does this result fall out of the definition? How do other corpuses stack up when this analysis is performed?