| If you have the chance, try the following: 1. Find a paper you like in a field you want to learn. 2. Use the keyword filters to filter down to papers that match your criteria. 3. Add a bunch of the interesting ones to a new graph using the purple "+" buttons. 4. On the "new" graph page, check out the similar papers section. If any of them are interesting, add those to the graph. 5. Repeat until you don't find anything else that is interesting. The similar papers section uses a link prediction algorithm that basically says, if two papers cite a bunch of the same papers, rank them higher BUT if the paper they cite, is cited a bunch of times, don't give that connection much weight. The net effect of this is that it doesn't really matter if the paper was highly cited, only that it cites the same niche of papers as the ones you just chose. Also, because of the temporal nature of academic literature, the papers it brings up tend to be the newer and harder to discover papers. The results are pretty great and it's as close to the "serendipity" slider that you'll get right now. EDIT: Formatting |