Years ago, I started referring to search results as recommendations, seeing how they’ve been starting to look more and more like that part of a page at Amazon that says “people who viewed this book also looked at these books.”
When someone searches at a search engine, one of the things they look for in the search results they receive are trustworthy pages (or recommendations) that look (and are) legitimate. How does a search engine deliver pages that are trustworthy?
One way to do that might be to try to boost pages in search results that the search engine feels are more trustworthy – and Google developed a version of Trust Rank to do that with. The inventor of Google’s Trust Rank (which differs from the version that Yahoo invented) is Ramanathan Guha.
In addition to being one of the major forces behind custom search engines at Google, he authored a paper on Trust and Web pages while he was at IBM before he joined Google, which makes for interesting reading – it’s at Propagation of Trust and Distrust (pdf).
Author Badges and Trust
A couple of years ago, I wrote the post, After Authorship Markup, Will Google Give Us Author Badges Too? It’s about how Google might display within search results some signs that an author is the creator of content, including a link to their Google Profile that potential readers can use to verify the author is whom they claim to be.
Google has been showing author profile pictures for with some search results when a searcher wasn’t logged into their Google account, leading many to the conclusion that Google might implement an author rank as some point, where pages with rich snippets showing these authors’ photos might be the kind of pages that get boosted in search results.
With most rich snippets, the addition of these profile pictures has been cited as a visual change that breaks up a traditional looking 10 blue links set of search results, and a possibility that the profile picture might influence more click throughs. The Author Badges patent doesn’t promise the display of a search result with a photo in it, but it does mention the possibility of including a byline that is attached to a link – that a searcher can use to learn more about the author by clicking upon it. It seems like the intention behind that is to improve trust in those results.
Google Webmaster Evangelist John Mueller announced at the end of this past June that Google was redesigning these author badge results so the they won’t display photos or circle counts any more, but would continue to include bylines. I’m working on another post presently that explores changes in search results based upon Google+, and it will look at some Google studies that may have influenced changes to those in the past.
It’s possible that Google is still using their version of Trust Rank, which relies upon the annotations or labels that people create for pages in their Custom Search Engines, and may boost pages that have been labeled with something relevant to the queries used to find those pages. Those Authorship bylines may continue to appear in search results for searchers to use to verify the identity of the creators of pages.
Or, if they were available, you could just place “Trust Buttons” on people’s websites that people can click on to tell Google that they trust the author of the Web site.
A patent granted to Ramanathan Guha last week describes how Trust buttons mght work at Google to provide recommendations from trusted sources. It’s possible that Google may never develop trust buttons. But, with things like unique digital signatures on Google Accounts, it is possible for Google to follow and base trust recommendations on people who produce content for the web, and who search and may leave a search history behind that could be relied upon by others who may indicate that they trust those people.
Trust buttons would be one way of obtaining recommendations from someone whom you know and trust.
Asking for recommendations from someone whom you’ve identified as being a person you trust might lead to direct recommendations, or those recommendations could influence search results that you see.
The patent tells us that identifying that a searcher trusts recommendations associated with a specific entity can involve:
- Receiving input directly from the searcher indicating the user trusts recommendations associated with the entity (such as in the pressing of a “trust” button – though an actual button might not be used);
- Using a web crawler to examine web pages to locate information indicating that the user trusts recommendations associated with the entity;
- Examining web visitation patterns of the user and inferring from the web visitation patterns that the user trusts recommendations associated with the entity; or
- Examining computer-based records belonging to the user, such as an email contact list or an instant-messaging chat list, and inferring from the computer-based records that the user trusts recommendations associated with the entity.
The patent is:
Method and apparatus for obtaining recommendations from trusted sources
Invented by Ramanathan V. Guha
Assigned to Google
US Patent 8,762,394
Granted June 24, 2014
Filed September 13, 2012
One embodiment of the present invention comprises a system that provides recommendations to a user.
During operation, the system performs a computer-based lookup operation to identify trusted entities that the user trusts (or is likely to trust) to make recommendations.
Next, the system performs another computer-based lookup operation to obtain recommendations associated with the identified trusted entities. The system then uses the obtained recommendations to provide recommendations to the user.
In a variation on this embodiment, the system additionally receives information indicating that the user trusts recommendations associated with an entity. The system then stores a corresponding record for the trust relationship between the user and the trusted entity in a computer-based lookup structure to facilitate subsequently identifying trusted entities that the user trusts to make recommendations.
The patent describes attempting to understand who we trust, and it might limit the topics where we seem to trust a person to certain things, like SEO or medicine or soccer or physics.
An actual “trust” button might not be involved, but if you frequently visit the website of someone, follow links that they post on the web, and display other behavior that appears to indicate some level of trust, it could potentially result in recommendations made to you based upon that entities actions, or pages in search results boosted based upon pages they might be affiliated with, or that they’ve given attention to.
A trust relationship might “decay” if it seems like the trusting user isn’t taking actions on the recommendations from the trusted entity.
Entities trusted by the trusted entity might be considered transitive trusted entities by this system, and could be trusted by the searcher/user trusting the original entity.
We’ve heard the phrase “trustrank” used before in relation to Yahoo’s trustrank algorithm, which looks at how many links away that pages might be from a seed set of “trusted” or “good” pages.
Instead of just limiting trust to a link analysis, this trust button approach looks at the relationships between people (and likely organizations or businesses as well, since they are entities, too) creating content, making recommendations, and visiting web pages, and people who might find those recommendations and that content useful and helpful, and something they might trust.