Why aren't some articles linked to the next one?
This can happen for a number of reasons, the most common one being that the link to the next article is either in some unprocessed info-box of the page or in the depths of the note section. Either way, opening the original article on Wikipedia should reveal the location of the link.
Why does swapping search terms produce different results?
Wikipedia is a “directed graph”. That means that article A might link to article B, but not vice versa. For instance, the article on Poughkeepsie links to the article on the United States, but the article on the US doesn’t mention Poughkeepsie. This explains why sometimes, when a WikiBinge path can’t be generated, swapping the terms can bring some luck.
How many articles and connections are used for WikiBinge?
Roughly 3 million articles with 40 million connections.
What happens when I ♥ a path?
It enters into limbo before being reviewed. The most beautiful paths are added to the random collection.
Will this be available in other languages?
If you wish to see WikiBinge in another language, contact me. With enough time, resources and requests, it can be done!
How was this done?
The original approach to "finding the most interesting path" was to compute the Rich Club of the Wiki-graph and prune it from the network. However, dissatisfied with results and not comfortable having to draw a line for "Richness", a more scalable approach was discovered: compute PageRank on the Wiki-graph and assign as weight of each edge the pagerank value of the destination node. A WikiBinge path is then simply a shortest path using these weights: the algorithm will then favour paths passing through articles with lower PageRank values. The underdogs are the kings of WikiBinge.
What tools did you use to make this?
A Wikipedia dump, igraph and some code.
//TODO list is already filled with proposals! Show an infobox with stats on each Wikibinge (such as categories used, length of path, etc). Let users pick the routing strategy (lower pagerank weight, shortest path, avoid subgraphs by category, etc). If you got some interesting idea, contact me.