Not so long ago, we had the opportunity to observe the algorithm of the social network Twitter, which can predict the appearance of trend topics on the network before they appear. Now, the developers have rolled out a new algorithm that allows them to separate the true tweets from the false ones, which can be very useful, for example, in emergency situations.
The self-learning system uses a number of parameters that reveal the level of confidence in the information in the tweet. For example, if it comes from an account with a large number of followers, if it contains a link, or if it is expressed in a negative tone, it is more likely that it will be true.
Now the algorithm determines the truth of the tweet with a probability of 86%, which is much higher than, say 1 to 1. Therefore, it can be run. Of course, there is plenty of room for improvement in the future, but such software can run much faster than people, and therefore, in some cases, it can help more effectively.
By the way, if you're wondering how Facebook and Twitter fought for Instagram, enjoy reading.
Source: theverge.com
The article is based on materials .
Comments
Post a Comment