A Guide to The Contextual Web
It's the end of 2008 and everyone on the Web is hurting due to the economy. But we know that things will get better, because slow-downs eventually bury the old and give birth to new evolutionary ways of doing things.
One of these evolutions started quietly in 2008. We are witnessing the rise of a new kind of web: contextual. You might not have heard or thought about it much yet, but you are already using it today. Search remains the killer app on the web, but context is quickly become a viable contender. Why? Because context is what happens instead of search.
What Is The Contextual Web?
Until recently on the web, most sites have not been software - only data, a bunch of flat HTML pages. The software that you used to look at them was, of course, the web browser. The problem was that the browser had no idea what the pages contained, and it did not know what you were doing. Because the browser could not infer your context, it could not help you explore related and relevant information.
The contextual web experience is fundamentally different because there is an understanding of what the user is doing. The combination of the information on the page and the user's behavior creates the context. Once you understand the user's context, you can be more helpful. So, contextual technologies have the potential to bite into the pie that today belongs to search, because it is able to bypass search.
Consider the difference between looking at the IBM home page and a movie page on Netflix. The context is very different, and so likely is the user's intention. The user may be looking at IBM because she is looking for a job or researching IBM products. When the user is looking up a movie, it is because she is thinking about renting it.
The Key Properties of The Contextual Web
The Contextual web will happen when browsers and websites evolve to recognize what users are trying to do. It is the web with less choice and more meaning, where instead of Googling all the time, we Google once and then the rest of the information is available to us automatically, based on our current context.
Here are the key properties of the contextual web experience:
- Relevancy: understanding the user's context better drives content relevancy.
- Shortcuts: contextual shortcuts reduce the need for raw search.
- Personalization: context is based on user intentions and history.
- Remixing: relevant information from around the web is instantly available.
What are some specific examples of contextual technologies that are improving our web experience right now?
Markup Technologies
One of the keys to inferring user context is understanding the underlying information that the user is looking at. This is why the contextual web is related to -- and, to be more precise, is powered by -- semantic web. We have written a lot here on ReadWriteWeb about semantic technologies. Notably, we discussed the difference between the top-down and bottom-up approaches to semantic web, both of which are important for understanding the contextual web.
The bottom-up approach to context is about annotating pages. For example, all modern browsers can detect if the page you are looking at contains an RSS feed. This happens because the browser looks for a tag in the head of the page, which declares type
Another form of markup that has been discussed recently is microformats, which offer an XHTML-compliant way of embedding metadata about people, places, events, and reviews in existing web pages. Even though microformats are not ubiquitous today, there are clear benefits to using them. The image below is from an excellent post by Mozilla UI lead Alex Faaborg about leveraging microformats in the browser.
Building on the hAtom microformat are Web Slices, introduced by Microsoft in Internet Explorer 8. Web Slices enable publishers to notify users when the information in their web pages changes. For example, Weather.com can create a Web Slice that tells the user when a local weather update is available. eBay can deliver a Web Slice that notifies the user when the price of an auction changes. Similar in concept to RSS, Web Slices focus on updates in part of the web page, enabling publishers and users to communicate directly via the browser.
There are other markup formats that help provide context. For example, popular add-on Cooliris offers a markup format for signaling that a site contains images. By placing a bit of XML code in their home directory, site owners enable users to experience their images using the stunning 3D visualization developed by Cooliris. Another markup format, developed by AdaptiveBlue [disclosure: this is the company I founded], is called ABMeta. This format allows publishers to annotate pages that contain information about books, music, movies, wine, restaurants, stocks, and other everyday things.
All of these markup-based approaches face the same issue: publishers have to do the work of actually annotating the pages. And while the semantic web community has been very vocal about the benefits of annotation, the majority of the web is still flat HTML.
Page 2: Widgets
Comments
Subscribe to comments for this post OR Subscribe to comments for all ReadWriteWeb posts
Hi Alex, too bad Juice (covered on RRW earlier) didn't make the list, as I think it's quite competitive with the other companies listed. Anyway, just chiming in on an otherwise excellent article on the state of the Semantic Web.
We have some new (intelligent discovery) tricks up our sleeve for 2009, that we feel will bring another exciting contextual layer to your browser. The problem we're aiming to tackle is how to make "contextual power" part of your regular browsing experience, without disrupting that same experience.
Posted by: Thijs | December 22, 2008 9:44 PM
Problem: These pieces do not add all up to a coherent whole; there are some good add-ons and such, but the hype of semantic has far outpaces the actual delivery of the implied promise. I'm all for being optimistic about cutting edge, but I have yet to see a real, can't do without semantic application or web extension that I just can't do without. They have all been interesting, but not compelling.
Maybe the real breakthrough will be when I'm wowed by a web application, and someone has to tell me that some clever use of RDF and OWL made this all possible. Having read the specs, and also having been monitoring the hypeosphere for going on five years, and having worked for a semantic web tools company, I hope that the big thinkers lay ground work for real stuff we can use.
Posted by: Alan Wilensky | December 22, 2008 10:15 PM
Wonderful add-on descriptions, thank you. I deduced some time ago that browsing FF extensions is like going down an exceptionally long rabbit hole, so I tend to find them only via recommendation these days.
Pondering: one individual's web experience is so different from another's, and so different across (his own) uses... do you foresee the "contextual web" being sometimes more of a hindrance than a help in the nearer future, before this technology comes to fruition?
And when it does, it will sure be interesting to see how the web world reacts to (truly) uncannily well-targeted advertising.
Posted by: krgaskins | December 22, 2008 11:28 PM
Great summary of extremely interesting space. Good job Alex!
I think the most important point to note in the whole article is that contextual technologies can remove need for search in certain situations.
You can also look at contextual technologies as search done differently (not through keywords).
I also agree with Alan's comment that most of the solutions are hype and that only a few deliver true value. Test of time will show which ones.
I'd add another one to the list - Aka aki. Quite local to Germany, but definitely showing where the contextual technologies are going in mobile.
I am looking at all this developments as the groundwork for the new paradigm of how computers are going to help in the future.
Andraz Tori, Zemanta
Posted by: Andraz Tori | December 23, 2008 12:41 AM
Semantic web and context will replace search, then Google will be facing problems.
Posted by: LEADSExplorer | December 23, 2008 2:00 AM
Disclosure: Our company is very interested in building smarter, better links, but we tend to think of the space as "Discovery" instead of "Contextual".
A drawback of the many of the existing services is they don't deliver new information to the user, or provide substantially more value than opening new tab.
- "Look up this term on Wikipedia!"
- "Buy on Amazon!"
- "Preview this page in a hover!"
etc.
There's no new information in any of these & it strikes me as a lot of "overhead" for saving an infrequent search on wiki/amazon/netflix etc. AFAIC, the "Preview this site" concept died with "Open in new tab".
On the other hand, those who do offer NEW information and surface data that was otherwise unavailable will enjoy the most success. i.e. "Discover" things you didn't know to search for.
And by the way, that's exactly why I love your posts--thanks to your links I always discover new companies and ideas I didn't know to look for by myself.
So there's no doubt that something like context/discovery/sematic etc. will eventually become a big space. And we can certainly enjoy and try to learn from the many interestesting approaches along the way. Thanks for posting.
Posted by: Israel LHeureux | December 23, 2008 2:26 AM
Well written article but I am surprised to see the glaring omission of LOCATION - quite possibly the most important context of all. From the post:
The Key Properties of The Contextual Web: Relevancy, Shortcuts, Personalization and Remixing...
OMFG location is not listed? WTF?
Here's hoping everyone at RWW will wake up and shake off this "chain to the desk" view of the world.
Posted by: Todd | December 23, 2008 4:07 AM
I agree about Todd's comment about location being a key element. Already some standards are emerging around providing location from the client to the server which will make location based applications much easier for developers.
I also think that personalization, while mentioned, isn't focused upon enough. I believe we only tap into a small percentage of the information we know about a person (their browsing history, likes, dislikes, etc) while serving them up the web. Different types of AI can be used to make inferences based on data and improve these guesses over time.
Posted by: Frank | December 23, 2008 4:49 AM
Informative, dense & thoughtful piece. Interesting no mention of how/where/whether Chrome or Safari are also driving in this same functional direction. A post-search world --- hard to fathom, but provocative.
Posted by: Thom Kennon | December 23, 2008 5:36 AM
Fascinating read
Isn't contextual just another way of saying that the search and the browser integration are the same thing, like a marriage between the two. A lot of the big players like Google and FaceBook are doing this contextual content rendering along with intuitive Browsers that have social integration that are contextual, for instance Delicious and Diigo are basing their browser extensions on users actions thus the contextual element is used again.
It's an ever increasing intelligent or as you say contextual web that is driving the innovation and along with semantic attributes we are in for a richer browsing experience. One thing’s for sure the big players will reap the reward. I also believe that everyday users like bloggers and active social network users can tap into this vast oasis and better serve their visitors.
Thanks' for the read
Posted by: Bill Masson (WWAH) | December 23, 2008 5:37 AM
Alex,
In a nutshell, I assume you are trying to unveil an Web incarnation that uses context to facilitate:
1. Serendipitous discovery of relevant "Things" (moving from SEO to SDQ [1] )
2. Less subjectivity and more objectivity
3. Less obtrusive style of advertising (i.e., we will soon be able to actualy read Web content from traditional media behemoths without those poorly located Ads)
4. Expose pathways to other potentially relevant things (i.e., HTTP based Data Source Names aka. URIs, will not be obscured by opaque Web pages and Web Services)
Links:
1. http://tinyurl.com/9ny5qx -- post about Serendipitous Discovery Quotient (SDQ)
Kingsley
Posted by: Kingsley Idehen | December 23, 2008 6:05 AM
I knew that above comment was K. Idehen before I even read the signature line!!! I knew it!
Posted by: Alan Wilensky | December 23, 2008 6:10 AM
Agree that contextual is not the right word, we're talking about semantic web still.
Regardless of how effective they are, plugins are fine for early adopters, that's about it.
Ubiquity is going to be great, kind of like moving Spotlight to the web.
Posted by: Dave Evans
|
December 23, 2008 6:19 AM
Actually web pages are not data, they are human-readable information. Data is the in the more or less inaccessible databases that lie behind the websites.
When web pages stop being static, much of the inherent value of the web as a set of indices vanishes. Search engines lose their relevance when pretty much anything can wind up on a given URL. The HTTP protocol also has well known problems with managing state, and the various standard approaches to dealing with this problem are all inadequate in various ways.
It's certainly the case that providing dynamic web page content, backed by some understanding of what the user is doing, will increase the value of a given user interaction, but doing this pervasively will also diminish the value of the web itself.
So it's a trade-off. Providing "contextual" or "dynamic" or "personalized" or "customized" (whatever your current buzz-phrase is) info is great, but it has its place, and shouldn't be a universal approach to site design and implementation.
Posted by: Miramon | December 23, 2008 8:45 AM
Great article. I think content has been dethroned, context is the new king!
Quite surprised to see no mention of APML under markup though, this is game changing in the search for context IMHO.
angus
Posted by: Angus | December 23, 2008 9:02 AM
Miramon makes a very important point, which is how we deal with the bridge between current methods & technologies for discovery (search) and the types of content we find and contextually consume.
This gets at the - so far - enduring & painful tension between SEO friendly web pages and the types of user engagement realized by enriching contextual content, data & experience.
So we either wait for the search bot algos to keep apace or we run out ahead and wait for them on the frontier.
When we turn to the grid humans want discovery, experience and distribution interlaced and seamless. And, perhaps, to the point of the original piece - the browser seems to be the medium for this delivery. But we need to figure this out people...I am NOT getting any younger.
And I want it too.
Posted by: Thom Kennon | December 23, 2008 11:49 AM
thanks..
Posted by: dairesel kapı | December 23, 2008 12:21 PM
I'm sorry, I don't see how you can speak of discovering relevancy of unknown content from local context when it follows that all that can be possible to "discover" is more of the same (things already in local context)- spread out over the web. That does not lead one to what ought to be relevant. It only leads to what is already been judged only possibly relevant in the past.
Like when you are just exploring, as on a tangent, that turns out to be a dead-end; all that is gathered up in what these programs "infer" from the data -- to use Alex's word. It is also used in the heuristic -- or blind guessing -- at what is relevant. What utility that?
Yet what a parade of products there are. Any of them are apt to gather the notions of what is relevant to you from your own bookmarks, from the messages in your in-box, from all your devices. I have to admit; it is smart. It may surprise some people to find out what all those notions are and how they possibly fit into their personal framework. But all of those notions combined or "inferred" are not the substance of relevance, nor are they the sum total of all that is relevant. They are only some artifacts littering one's environment.
So, imo, people should lower any expectations that these products are going to discover what ought to be relevant. And I would caution anyone to dispel any notion that these products have any sort of power of inference, within their programs. People use their powers of inference to judge when something is relevant *especially when* it does not appear to match any previously gathered facts, matters or artifacts in their existence.
I believe producers have an obligation to consumers to be careful not to confuse machine inference from data, for the human inference and induction of the substance of meaning and relevance.
-Ken Ewell
Posted by: Ken Ewell | December 23, 2008 12:26 PM
Very nice overview, Alex. A couple more plugins:
http://headup.com
http://mashlogic.com
Posted by: Gray Norton | December 23, 2008 10:17 PM
Excellent article. One minor suggestion, search can be filtered based on context...so sort of limiting to say context can bypass search.
@Alan - Contextual relevance can be achieved when we can infer intent of the user. And we all know how notoriously difficult it is to infer intent of humans. This is not a trivial problem, hence the lack of full-blown applications and experiences. However the pieces are falling in place and like any complex technology it is by many false starts we truly make progress.
@Leadsexplorer - Google wont be replaced. In fact they can provide more targetted results by using context.
@Todd - Location is indeed a key contextual factor
@Ken - You bring up a valid point. People determine relevance not machines. However there are limited use cases where machines can simplify determining contextual relevance.
One factor that needs to be emphasized is this. Once contextual relevance is established the sum of actions that can be layered on them is where effort needs to be spent.
Posted by: Mahesh CR
|
December 24, 2008 1:24 AM
I think this is a nice collection of widgets and applications that add information to web pages based on their content, but I don't see any algorithms here that are about to make the 'contextual' or even the 'semantic' web relevant.
Surely 'context' needs to span more than just additional information and extra links to similar articles? (I can get that all from Google as it is...) Surely true 'context' would require a history, stored and made available to all applications? Even then, when engaging with the web and trying to find something outside of what all these widgets and apps think is your 'context' is going to be really annoying. I see no need for the web to second guess my intentions.
"this [web] understands what we are doing and helps us"
This is nonsense. Other than meta information, this article is still about search. There is no way the collection of widgets and apps mentioned here are going to 'understand' what we are doing. Search, whether you call it that or not is here to stay and my guess is Google is going to lead it for some time to come.
Posted by: joedamage | December 24, 2008 3:04 AM
I see the value of defining data for users and browsers but without a universal markup definition will it ever build up momentum?
Posted by: Jon | December 24, 2008 8:52 AM
Derek,
Here are Aldo's points in a nutshell:
1. XBRL is an XML based notation for expressing Financial Reporting Data
2. XBRL is a representation of Data
3. RDF is a framework in which all Data items are Resources (i.e. useful things that have Identity) or Data Objects (or Entities) and these Items/Objects/Entities create a "Mesh" (a Graph) based on their Attribute values and/or Relationships with other Items/Objects/Entities.
XBRL instance data can be presented in RDF Linked Data form such that each item in the XBRL report becomes an Entity endowed with an Identifier.
In the case of Linked Data, the Identifiers incorporate the HTTP protocol such that each Object/Item/Entity is de-referencable over an HTTP network ( which does not always mean the public World Wide Web).
Linked Data is simply about a continuation of the long desired pursuit of platform independent distributed data objects.
Back to XBRL, we already convert XBRL instance docs (XML) into RDF by the use of RDFizer or RDFization middleware. What this means is that right now we can already build Entity/Object graphs from any XBRL document "on the fly" so that you can explore XBRL data object in granular form.
Some live examples:
1. http://demo.openlinksw.com/about/html/http://www.sec.gov/Archives/edgar/data/40545/000004054508000065/ge-20080930.xml%23GE_CONS_3Q08 - Part of GE's XBRL Instance Data in RDF Linked Data form
2. http://dbpedia.org/resource/Linked_Data - About Linked Data from DBpedia (which is an RDFization of Wikipedia)
I hope this clear up matters. Yes, as you say, you can "Mash" when dealing with XML and the rest of the Web 2.0 stack. Just bear in mind that Linked Data is simply about adding the ability to "Mesh" to the mix since the data is structured at a more granular level.
Kingsley
Posted by: Kingsley Idehen | December 24, 2008 8:59 AM
ı have followed your writing for a long time.really you have given very successful information.
Posted by: oyun | December 24, 2008 4:41 PM
Excellent. Good reading. Great explanation of a complex, and often vague, subject.
Posted by: Chris Loft | December 24, 2008 6:26 PM
.What a post!! Very informative and easy to understand.looking for more such posts..Lets share some information regarding the technology.
Posted by: webform | December 25, 2008 12:47 PM
Please enter on http://VideoclipManele.blogspot.com
Thank you very much from Romania [Europe]!
Posted by: Sakin1985 | December 26, 2008 6:53 AM
Nice Post. Very informative. Thanks a bunch!
Posted by: Timothy | December 29, 2008 7:48 AM
I hope this clear up matters. Yes, as you say, you can "Mash" when dealing with XML and the rest of the Web 2.0 stack. Just bear in mind that
Posted by: araba yarışı | January 7, 2009 12:25 PM
I hope this clear up matters. Yes, as you say, you can "Mash" when dealing
Posted by: futbol oyunları | January 7, 2009 12:26 PM
I hope this clear up matters. Yes, as you say, you can
Posted by: motor oyunları | January 7, 2009 12:27 PM
One of these evolutions started quietly in 2008. We are witnessing the rise of a new kind of web: contextual. You might not have heard or thought about it much yet, but you are already using it today. Search remains the killer app on the web, but context is quickly become a viable contender. Why? Because context is what happens instead of search.
Posted by: yemek oyunları | January 7, 2009 12:28 PM
One of these evolutions started quietly in 2008. We are witnessing the rise of a new kind of web: contextual. You mi
Posted by: araba oyunları | January 9, 2009 10:42 AM
thanksssssssss
Posted by: Savas Oyunları | January 15, 2009 9:34 AM
thank you
Posted by: Yeni Oyunlar | January 15, 2009 9:35 AM
:)
Posted by: 3d Oyunlar | January 15, 2009 9:38 AM
sssssssss
Posted by: Show Oyun | January 15, 2009 9:39 AM
teşekkürler...
Posted by: Zombi Oyunları | January 15, 2009 9:40 AM
thank you very muchh
Posted by: Oyunlar1 | January 15, 2009 9:43 AM
thank you admin
Posted by: Maç Oyunları | January 15, 2009 9:45 AM