Web graph compression by edge elimination

Summary form only given. This work focuses on the problem of compressing the Web graph by means of eliminating some of the edges in its link structure. An algorithm is used to divide the task so that it can be executed on parallel processors. Ran on a test bed of generated Web graphs, the algorithm improved the compression ratio of both Huffman-coding schemes and Adler and Mitzenmacher's find reference algorithm tangibly. In the find reference case, improvement was up to 90%