Results (
Thai) 1:
[Copy]Copied!
Fig. 18.1: Functional components of a typical web search engine. A page, pi , is located on the web by the crawler and its content, the terms t1,...,tn, are retrieved and indexed as part of an offline process. In response to a search query, the engine probes the index to retrieve results which match the query terms, pi,..., pj, which are then ranked by their relevance according to the search engines particular ranking metrics, before being presented to the searcher as a result-list.Improving the ranking of search results became the challenge for these early search engines and even the race for the largest search index took a back seat in the face of this more pressing need. It soon became clear, however, that relying solely on the terms in a page was not going to be sufficient, no matter how much time was invested in tweaking these early ranking algorithms. Simply put, there were lots of pages that scored equally well when it came to counting matching query and page terms, but few of these pages turned out to be truly relevant and authoritative. Although term matching information had a role to play in overall relevance, on its own it was insufficient, and it was clear that there was vital information missing from the ranking process.The missing ingredient came about as a result of research undertaken by a number of groups during the mid 1990’s. This included the work of John Kleinberg [40] and, most famously, the work of Google founders Larry Page and Sergey Brin [13].
Being translated, please wait..
