A Study About Ranking Algorithms Of The Search Engine

The emergence and rapid development of the Internet to make information retrieval environment have undergone significant changes in the skyrocketing number of pages, so that people accurate access to information more difficult, in this context, search engines become an indispensable information retrieval tool for people , but the number of pages returned by search engines is often massive, and how to get users to mass speedy return of results to find the most accurate information, it is particularly important. The search engine's website relevance sorting algorithm, we can determine whether the user can find the top-ranking web pages you want information, so sorting algorithms search engines use a direct impact on experience, determine the usefulness of search engines. Existing search engine ranking algorithms, web-based link structure of the algorithm is based, the main two kinds of representation of the algorithm is PageRank algorithm and the HITS algorithm, based on these two algorithms many scholars and research institutions at home and abroad has conducted a new exploration and improvement.This paper analyzes the development of search engines at home and abroad, on this basis, the classical sorting algorithms in-depth analysis (such as PageRank and HITS), will be at home and abroad to improve the existing algorithms are summarized and reviewed, and for the ARC algorithm (the improved HITS algorithm) the characteristics and shortcomings, propose link-based similarity of the improved algorithm, and use Bayesian probability model to derive simplified method. Then in the establishment of search engine experimental platform, the link-based similarity of the ARC algorithm is verified experimental results show that the improved algorithm is effective control of the theme ARC drift, improve search engine performance.