We developed an algorithm to predict which version of the "same" announcement an author should choose to get more attention. The main idea is to automatically learn what wording works better by examining a large number of pairs of tweets posted by the same author containing the same url.It's worth noting the comparison tool uses a simplified version of the algorithm.
I decided to look at some geospatial news tweets from last week. The pairs were not from the same people, but did refer to the same content, a press release.
