Abstract | Searchius is a collaborative search engine that produces
search results based solely on user provided web-related
data. We discuss the architecture of this system and how
it compares to current state-of-the-art search engines. We
show that the global users¢ preference over pages can be
efficiently used as a metric of page quality, and that the
inherent organization of the collected data can be used to
discover related URLs. We also conduct an extensive experimental
study, based on the web related data of 36483
users, to analyze the qualitative and quantitative characteristics
of user collected URL collections, to investigate how
well the users URL collections cover the web and discover
the characteristics that affect the quality of the search results
under the proposed setting. |