Francisca Varela

The Web structure can be modeled as a directed graph, where the pages are nodes and the links between pages correspond to edges. These Web graphs are widely used for analyzing the Web, for example, for the calculation of Page-Rank and for spam detection, among others. The size of these graphs makes their manipulation in main memory unfeasible. In 2005, the graph representing the public static Web was estimated to have 11.5 billion nodes, and 15 edges per node on average; this requires around 600 GB in its plain representation. Hence the motivation for this work, which consists in building compressed data structures for Web graphs. A compressed data structure aims to represent data using little space, in main memory or on disk, while supporting queries without decompressing the whole data. The main advantage of these structures is that more data can be kept in main memory, or fewer disk transfers are required. This is of utmost importance, since an access to disk can be a million tim...