Data Mining: Know It All

Chapter 10. Social Network Analysis

The size of the Web and the reach of search engines were both increasing rapidly by late 1996, but there was growing frustration with traditional information retrieval (IR) systems applied to Web data. IR systems work with finite document collections, and the worth of a document with regard to a query is intrinsic to the document. Documents are self-contained units and are generally descriptive and truthful about their contents.

In contrast, the Web resembles an indefinitely growing and shifting universe. Recall, an important notion in classic IR, has relatively little meaning for the Web; in fact, we cannot even measure recall because we can never collect a complete snapshot of the Web. Most Web search engines ...