- November 2009 (1)
- May 2009 (1)
- April 2009 (1)
- February 2009 (1)
- December 2008 (1)
IVT News
Google search result personalisation goes mainstream
Fri, Dec 12 2008
By Jonathan Oxer, IVT Technical Director
I've been blabbing on for years about how eventually search engine results will be personalised so that when you do a search the results you get may be totally different to the results of someone else doing the exact same search. The basic principle is that everyone has different interests and requirements, so one person searching for "eagles" may be interested in birds while another is looking for information about the 70's rock band.
Google have experimented with a variety of personalisation techniques over the years but it's always been in special limited-test systems accessed through places like labs.google.com. A couple of weeks ago, however, they added personalisation features to the main public "www.google.com" search interface for the very first time - quite a historic moment, although it occurred without much fanfare! The event is indicative of a major change in the way search engines of the future will work.
To see the additional personalisation features you need to be logged in when you access Google. Up in the top right corner of the Google home page you'll see either "Sign in" if you haven't yet signed in, or your Google Account username (usually an email address) if you have.
When you perform a search you will notice that the results page may contain a couple of little icons next to each result: an up arrow that allows you to "promote" that particular result, and a cross that allows you to hide it. In effect it's a bit like a "thumbs up / thumbs down" voting system for results so that you can blacklist rubbish that comes up in the results page. At present the votes you submit only apply to your own results: one user voting up a site won't cause it to rise in the search results for all other users. However, I'm sure that Google will be amassing a wealth of data as people start using the system and it may well be factored into future updates to the PageRank algorithm.
You may also notice that the search results page says "Personalized based on your web history" near the top right. You can click the little "More details" link to find out exactly how your results were modified, typically based on previous behaviour such as which sites you clicked on in previous searches. Sites that you click on more often in search results will tend to appear higher in results for future searches.
It's certainly not Earth-shattering from a functionality point of view, but it's definitely an indicator of the changing landscape with respect to search engine optimisation. The old days of search engines providing the exact same results to every user for any given keywords will soon be but a memory and everyone who wants to promote their site online will need to understand how search engines customise their results on a per-user basis.
Which reminds me, I'm well along the way with writing my next book, "Quickstart Guide to Search Engine Optimisation". It should be available in all good bookstores as well as online retailers such as Amazon and Barnes & Noble around February next year and will include a lot of behind-the-scenes information on how search engines actually work, so I'll let you know when it goes on sale.
Finally, a wrap-up of some recent events to end the year.
In the last eBusiness News I asked for feedback about my tech predictions for 2009. I received more email about that than any other topic I've ever written about, and it was also picked up by Fairfax Media and used as the basis of a story that was the lead item in the technology sections of The Age and Sydney Morning Herald online for three days in a row. I had email responses from all over the world including from an old friend living in China who was good enough to provide his first-hand experiences with Internet censorship.
His comments were pretty much what I expected: attempting to censor the Internet can be effective against casual users who can't be bothered working around it, but to anyone who really wants to get around it they can using encrypted email, secure tunnels, The Onion Router project, and a multitude of other techniques. It's therefore an annoyance to regular users while not stopping those who it's actually meant to block! If it's ever applied in Australia it would be a classic case of penalising the innocent majority in an ineffectual effort to stop the guilty minority. I'm sure the news over the last day or so about the discovery of a major pedophile ring including a police officer and a retired QC will be used as additional ammunition to support Internet censorship, but in fact it's a classic example of why Internet censorship is a total waste of time in trying to solve the problem - they just send DVDs to each other through the post! Spending money on Internet censorship in the expectation that it will prevent the spread of child pornography is totally misguided, and that money would be far better spent on investigation and enforcement.
I also promised that I'd give one of my books to the person who submitted the best response, and the winner was Paul Wayper from Sydney! Paul wrote an insightful email about the possible motivations behind the push for Internet censorship, so when I went up to Sydney last week to present a paper at the Open Source Developers Conference I packed his prize in my suitcase and delivered it to him personally.
Although it probably won't be of interest to you unless you happen to be into programming, you can now access the slides from my talks on Slideshare. My main presentation was on "Self Healing Databases", and you can see the slides at:
http://www.slideshare.net/jonoxer/selfhealing-databases-managing-schema-updates-in-the-field/
I also did two lightning-talks, which are talks with a 5 minute time limit. The first was about my recent project to mount a computer in my car and connect it to the Internet:
http://www.slideshare.net/jonoxer/geek-my-ride/
The second lighting talk was about a joke programming language I invented a couple of years ago and demonstrated running on the computer in my car to communicate with the engine management system:
http://www.slideshare.net/jonoxer/osdclang-on-mobile-devices
A couple of my developers also presented papers at the conference. Antoine Osanz did a very well-received presentation about efficiently representing hierarchical category structures in a relational database:
http://www.slideshare.net/ivtantoine/managing-category-structures-in-relational-databases-presentation
And Peter Serwylo presented a paper on why uptake of Open Source software is so slow among typical non-technical computer users, generating some interesting discussion among the audience.
Have a great Christmas / New Year break, and I'll be back in January with the next eBusiness News!
Cheers :-)
Jonathan Oxer
PS: Some of the projects we've completed recently at IVT include:
* Apollo Bikes (www.apollobikes.com)
* Association Solutions (www.associationsolutions.com.au)
* Connect Trades (www.connecttrades.com.au)
* E-lift (www.elift.com.au)
* Greenacres (www.greenacreslawn.com.au)
Learn more about these projects at www.ivt.com.au/portfolio
