About PCA News
PCA news shows a feed of ML-related arXiv papers that are getting traction on Twitter and the most interesting tweets that reference each paper.
Tweets are selected by a simple set of heuristics. First, I try to filter out tweets that simply include the paper title and authors. For the tweets that appear to have some original content, I estimate the author’s influence on ML Twitter by looking at the proportion of people I follow that follow the author. Then I add in the number of likes and retweets, with some adjustments to account for the the difficulty of getting engagement on tweets that buried further down within threads. Finally, I (recursively) add the scores of the all the replies to the tweet. This ensures that a reply will never appear without the context of the tweet it is in response to.Features under development now:
- Make past papers available and add a search based on paper title.
- Support blog posts from a whitelisted list of domains in addition to arXiv papers.