We tested our proposals using the Web graphs provided by the WebGraph project. We do not provide downloads for all the crawls, instead we would encourage you to cite their original source if you use them.
In particular, we tested the five crawls showed in the next table. The "bpe fast" column corresponds to the bits per edge required by the variant presented as Re-Pair CDict NoPtrs and the "bpe slow" corresponds to the one that mixes Re-Pair with Wavelet Trees. For more details about these variantes look into the documents section.
|Crawl||Nodes||Edges||Plain size (MB)||bpe fast||bpe slow|