The Hilltop algorithm is a ranking method from search engine research that scores pages according to the links of independent “expert pages”. The term is historical.
At a glance
- Type
- Ranking method from research (paper)
- Started
- 1999 (experiments in August 1999)
- Affects
- Scoring of target pages based on expert pages
- Confirmed by
- no, Google has not documented its use; adopted in 2003 according to reports
Last checked:
How does Hilltop work?
The method comes from a paper by Krishna Bharat and George Mihăilă, whose experiments date to 1999. It defines an expert page as a page created to direct people to resources, with links to many non-affiliated pages on a topic. Two hosts count as affiliated, for example, when they share the same first three octets of the IP address.
Hilltop works in two phases. In expert lookup, the system finds expert pages that match the query. In target ranking, pages receive scores when several independent experts link to them. If no suitable experts exist, Hilltop returns no results, because the authors tuned it for accuracy rather than query coverage. In the paper’s test, 2.5 million of about 140 million pages qualified as experts.
Hilltop and Google
According to reports, Google adopted the method in 2003. Google has not documented what role Hilltop plays in its ranking today.
Example
A consumer site lists providers of garden furniture and links to ten unrelated shops. If several such pages link to one shop, Hilltop treats that as a sign of the shop’s authority on the topic.
Sources
- Hilltop: A Search Engine based on Expert Documents – Bharat & Mihăilă (Univ. of Toronto)ftp.cs.toronto.edu
- Hilltop algorithm – Wikipedia (secondary source)en.wikipedia.org
- Google's Hilltop Algorithm – Search Engine Journal (secondary source)searchenginejournal.com
Last updated: