Yahoo Awarded Patent for Human-Enhanced Search Engine

Yahoo gets patent for human editor enhancements to search results.

Yahoo was awarded U.S. patent 7,599,911 today for “Method and apparatus for search ranking using human input and automated ranking”. The patent was filed in 2002.

Essentially, this method calculates search rankings based on both automated search algorithms and human editor input. In the patent, Yahoo describes a way that previously-collected input from human editors can be mixed in, or “blended”, with what its search algorithms return, resulting in better search results.

Yahoo describes the benefits and drawbacks of human editors:

Ranking by human editors reviewing search results provides more relevant ranking than automated processes and even search users, because human editors possess better intelligence than the best software and more clearly understand distinctions in pages, and human editors focus on areas of their expertise. For example, a human editor would more easily spot a page that is irrelevant but contains terms designed to get a high ranking from an automated process. However, human editors cannot process the volume of searches typically received by a search system and cannot keep up to date the queries they do process in view of the relevant pages that are added for consideration, modified or removed. In addition, in an open-ended query system, the number of possible queries can easily be in the millions. Even if editors concentrate only on the most common queries, the results change all the time as new data becomes available, old data becomes irrelevant, new meanings are created for old terms, or new events occur. If the results are based solely on what the human editors decided on one day, they might be stale and out of date the next day.

Yahoo proposes a way that human editors can adjust results for particular search queries:

Promotions and demotions might be absolute (“Rank this document first highest.”), relative to itself (“Rank this document four positions higher than it would otherwise be.”), or relative to another document (“Rank this document higher than this other document.”). Other types of promotion/demotion might include “remove this document from consideration no matter what the automatic system suggests”, “this set of documents are to be given equal (‘tied”) rankings”, “do not rank this document higher than position P” for some integer P, or the like.

A “blender” is then applied to automated search results and the editors’ inputs, resulting in a final ranking of search results.

An example in the patent is for the search query “medical conditions related to sports”. A human editor might lower the ranking of web sites for “tennis racket” in relation to web sites about “tennis elbow”.

Read Yahoo’s Patent (pdf).