Predicting Web Page Status

Abstract:     
The World Wide Web has become a key intermediary between producers and consumers of information. Web's linkage structure has been exploited by contemporary search engines to decrease the search cost for consumers while usually also rewarding the producers of higher status Web pages. In addition to influencing visibility and accessibility, in-links, as marks of recognition, accord status to a Web page. In this paper we show how Web page status may be predicted at least in part by page location and topic specificity. Moreover, we observe that the "philanthropic" contributions of a Web page, specifically contributions of information brokerage function, is also a good predictor of in-links. The observations are made in the presence of domain and topic-specific effects. Interestingly, all of these features that may predict status are "local" to a given Web page and within the control of the owner/author of the page. This is in contrast to the "global" nature of Web linkage-based metrics such as in-link count that are derived as a result of downloading and indexing billions of pages. Since the linkage structure of the Web has affects on browsing, crawling, and retrieval, our results have implications for vertical and general search, business intelligence, and content management.

Keywords: web search, search engine marketing, web visibility, status, influence

Working Paper Series