Automated ranking based on engagement data
The average paper pulls in over a 1000 articles, videos, photos per edition. Out of all that content, how does Paper.li decide which articles to present on a paper?
In order to unearth a good mix of engaging content based on all your sources, automated ranking evaluates articles against each other across all sources depending on engagement data.
For each type of content source, we've defined a set of social signals used to evaluate the "importance" of a shared article. The scoring of articles is based on three main elements:
- who shared the article
- who published the article: we attribute reputation scores to article sources, based on the monthly processing of 240 million articles
- the content of the article itself: we look at the article itself, taking into account when it was published, the language, topic covered.
We go back and fetch all tweets and content specified in your content sources, discard any duplicate URLs, URLs from private accounts or links that are broken, then analyze, organize and present the most relevant content (based on the top--max. 6--most relevant topics), video and photo excluded. To keep discovery relevant, we limit content to 250 articles, 100 images and 50 videos per paper.
If automated ranking is not working for you, you can change your ranking method to manual ranking, where you determine the source ranking yourself. Please see our support article, here.