Last Friday, Google published a new patent - Google's Agent Rank - to be included in their ranking system. This new concept is based on ideas discussed during a conference in 2005 by Ted Nelson, an information technology pioneer better known for envisioning the concept of "hypertext".
Google, like many other search engines, has been returning results based on the content of individual pages, with a rank assigned to each individual page. However, this new patent clearly states the intention to assign a new score depending on the reputation of each individual author. Thus, a blog, review page or article website could have different relevancy scores assigned to them depending on each author, or "agent" as it's referred to in the patent.
When a query is entered in Google, the search engine returns matching results based on query-dependent and query-independent data. The query-dependent criteria refers to those factors which are directly related to the query entered, such as keywords, keyword density, and so on. However, query-independent data classification, better known as page rank, reflects a series of other factors which are not directly related with the query, such as trust or authority based on the number of inbound links.
Google's Agent Rank Patent introduces a new tweak to the calculation of trustworthiness in query-independent data, whereby the authority of each individual agent, or content producer, is taken into account.
When content is produced by a single author, the relevance of that page will be calculated mainly on query-dependent criteria, by which the content of the page will have more importance. However, when more than a single author can be identified, each partition of the page will have a different relevance assigned depending also on query-independent criteria, such as the reputation of each individual author. The technical difficulty lies in identifying successful cases where different authors have contributed to an individual part of the content, and therefore in different levels of reputation having to be thrown into the mix.
In order to identify each individual level of authorship, the concept of digital signatures arises. New meta-data information could specify who the author, editor, publisher or reviewer of specific content is. Once the digital signature is available, individual scores could be assigned, therefore influencing the ranking results.
