For a long time web designers and content authors have been able to change what’s displayed on a page depending on who’s viewing it. However this is normally restricted to adverts and sidebars. Soon we could see that actual page contents adapting to suit the reader as well.
When you visit a webpage does it look the same each time you view it? Does it look the same as when your friend views it? Probably not and that’s because of personalisation and recommendations.
Unless it’s a very basic site the chances are that the page content changes depending on who’s viewing it and what they’ve viewed before. This could be driven by the site itself or from external feeds such as Goolge Adsense. It’s predominantly used to place adverts which are pinpointed to particular user types. The idea is that by targetting adverts to suit your preferences site owners will increase their revenue. By tracking which products and pages you view the site can build up a picture of what you prefer and will then push the appropriate banners at you. They do this using cookies which are small pieces of information stored by your browser.
When it’s done right, recommendations can be a very useful. Everyone accepts these days that if their going to use the internet they’re going to have to put up with adverts. So why not have adverts which are pinpointed at you and vaguely useful to you? Amazon of course is the master of this and by tracking your browsing habits are able to push their best selling items at you. This doesn’t always work, if you, shock horror, end up buying the product from someone else Amazon won’t realise and will still push it at you. Likewise buying a one off item, like a gift, will change your recommendations.
Recommendations tend to work primarily on advertising with very few people targetting the actual page content itself. However the next logical step could be to alter the main text on a page depending on who’s viewing it. As an example consider a Health care site offering advice on conditions. By analysing searches and pages views it should be able to build up a picture of the user even without them needing to login. Age and sex seem like fairly obvious ones to be able to establish. Knowing this information the page content could be refined to more closely suit the user. For example, the symptoms and treatments for a male patient could be displayed higher up the page if you know that your user is male. Likewise illnesses which are more serious for older people can display more prominent warnings depending on the age of the user.
Of course while this may all sound like a good idea but how do you actually do it in practise? The key to that is metadata and semantics.
According to this great post, Beyond Editing Web Pages, Into Creating Experiences, the age of the page is over. Nowadays contributors are being encouraged and in some case demanding to write information in smaller chunks rather than long pages. The theory being that the chunks can be combined dynamically at run time and presented to the user. Even this idea isn’t terribly new but up until now it’s been very hard to put into practise. The main problem has been that each chunk of text has required classification and tagging so that the system knows how to put the pages togeather. This metadata monster generally means that contribution is so complex to be unworkable. If you’re writing a page and have to tag each paragraph individually that’s a large commitment and one which most users would baulk at.
New technologies though are starting to change this. Things like Expert System and IBM Watson are capable of understanding documents. By using semantics they are able to infer the meaning within the the documents or paragraphs and are capable of categorising them automatically. Based on these categories it should be possible to amalgamate pages on the fly in response to user profiles. The age of the page may be over but the age of the truly dynamic web site may be just beginning.