Web searching

摘要

A human or hand-labeled ranking of URL results for a search query is compared against actual click data for the respective query/URL pairs (e.g., which URLs were actually clicked on by users when the URLs were presented to users when the search query was run in the real world). The human ranking or ordering of the URL results (e.g., pre-existing relevance ranking) for the query can then be adjusted, if necessary, based upon the real world click data (e.g., click relevance ranking). The modified pre-existing relevance ranking can be used in providing future search results.

主权项

1. A method, comprising:
computing a click relevance ranking of a first pair and a second pair based upon user log data, the first pair comprising a query and a first URL, the second pair comprising the query and a second URL, the computing comprising:
reducing click bias associated with at least one of the first pair or the second pair by determining a normalized click rate from the user log data;creating, after the reducing, a directed acyclic graph of a relevance relationship between the first pair and the second pair, the directed acyclic graph comprising a representation of at least one of the first URL or the second URL, the representation in the directed acyclic graph directly connected in the directed acyclic graph to a second representation of one or more URLs in the directed acyclic graph; andmapping the directed acyclic graph into a linear ordering to ascertain the click relevance ranking.