One strange and creepy patent
Imagine if you had a program that would automatically go through the posts on your site/blog and automatically find concepts and dynamically add links to related documents from the web. Now let’s go further and have that program track the people reading your page and by looking at their history and implicit past actions, ensure the links were tailored to each person reading said content. Wouldn’t that be freakin awesome?
Now imagine if Google wanted to do that. Not so cool anymore is it? Lol.
That’s precisely what was proposed in a recent patent awarded to Google. I really am not sure how it was intended to be used, but I’d have to imagine for their own properties… there would be wholesale mutiny if they tried using the toolbar (or Chrome browser) for such implementations. Here’s the patent;
Enhanced document browsing with automatically generated links to relevant information
Awarded; June 17, 2010 – Filed; February 23, 2010
Although this was an update to the one filed in 2004
Publication Date: 04/13/2010
Filing Date: 06/30/2004
How Google might add links to existing content
They give the example of an article/document about a hiker’s (Bill Cross) effort to climb Mt. Everest. The system would find additional links to supporting documents about Mount Everest, Novolog Peaks and Poles Challenge and maybe even reference something relating to diabetes (which he suffers from). Those skilled in the art, search geeks, get the idea I am sure.
Now, they don’t stop there, it gets personal. Let’s say for example the searcher/user has previously queried for items relating to photography. Other links can be included for items such as ‘Everest photography’ and so on.
The personalization data can also be geo targeted. Let’s say the user/viewer is from a given location. Then the system might also put links to localized relations such as ‘hiking Mt Rainer‘ for someone from the Seattle (USA) region. For those familiar with the various methods of personalization, I am sure you can extrapolate from there. We can further take this to include any form of personalization that they have at their disposal that are relevant to the document being viewed.
So it would be looking for related information to the document being viewed through query data and personalization.
“The information relating to the current document may include information based on the text of the current document. The text may be processed to obtain, for example: (1) all terms appearing more than some pre-determined number of times, (2) named entities that can be automatically extracted; (3) dates in the document; (4) author and publication names; and/or keyword or category extraction.”
The bits about named entities is interesting and sensible given whom worked on this patent. More on that later, but this is generally related to persons/place/thing. Terms that appear more than a given threshold are deemed to important and/or descriptive. The date element can be used in concert with these to find contemporary documents relating to the one being viewed. This would be especially important with News.
The question is; where would they use it?
The implementation is somewhat fuzzy though. The discuss displaying the notation separately, with a link and snippet, but also as an in-line link. A link that is added to the content dynamically. How this is proposed to be accomplished is unclear. Obviously Google can’t just start modifying the page without the page owners permission, thus the intent with this patent is somewhat unclear unless implemented on a Google owned property.
“(…) may incorporate the additional documents located by search component 410 into the current document (i.e., the document currently being viewed by the user) or into a new document that includes the current document. The additional documents may be incorporated with the current document in a manner that informs the user that the documents are available without being overly disruptive to the user’s reading of the current document. ”
Once more, that is about all they really discuss as far as implementation. There is no mention of 3rd part content nor use of a browser add-on nor toolbar. Where they were going this this isn’t entirely clear.
Now, beyond the tin foil of wondering where and how they were thinking of implementing it, this does make an interesting idea for anyone optimizing for search. Imagine a WordPress plugin that could do this for you? Sure, we wouldn’t have the level of personalization data that they do, but dynamically creating links within a document, say from top results for related terms, would be a handy gizmo and play well with trust algorithms such as TrustRank and HarmonicRank.
I leave you to ponder it all… just thought it was a strange one.