Some people argue that although the Semantic Web technologies are all in place, the world isn’t ready for the Semantic Web vision yet; that we’re still a couple of years away from fully embracing and adopting Semantic Web principles and attitudes.
Perhaps, though, the opposite is closer to the truth: that some of the biggest and best projects on the web today are based on Semantic Web principles - they just don’t happen to be using Semantic Web technologies.
It could be argued that Semantic Web models are exploding on to the web in spite of the technology, not because of it. Maybe we are using older, more limited technologies because of our lack of understanding and desire to learn complex new languages.
Current web applications that could be cited as Semantic Web proofs of concept are…
-
Google. The best discovery engine on the web relies on a single core principle to make it work: relationships between pages are valuable. This is more-or-less the exact same principle that the Semantic Web is based on - that relationships between chunks of data are useful.
At the moment, Google only uses anonymous relationships (why are two pages hyperlinked?) between two ‘blobs’ of data (each page ‘blob’ usually contains paragraphs of text, metadata, images and other information). Imagine the possibilities when, with the Semantic Web, the type of relationship is explicit, and the exact data points at each end are connected (for example, an image on web page A is defined as a photograph of the author of web page B).
-
RSS. These files provide a layer of data under the current web, that describe web pages and allow us to re-use and distribute these descriptions easily, aggregating the data as and when we need it (for example in our desktop RSS readers, or Technorati).
Guess what the main Semantic Web deliverable is? Pretty much the exact same thing - just a bit more flexible (we’re already seeing the boundaries of RSS being stretched as more and more is being asked of it).
-
Tagging. The need to describe individual items (blogs, photos, events) - rather than a whole web page - has led us (so far) to the ‘tagging’ microformat, where we re-use existing HTML markup to provide short keywords for items on a webpage.
As discussed above, this is precisely what the Semantic Web is designed for - describing chunks of information in a standardised, internet-friendly format.
-
Social Networking. One of the major web phenomenons of the last few years has been the growth and popularity of Social Networking sites and applications. All of these applications use the same main principle: each user creates a metadata record for themselves, recording data values for particular metadata properties (height, weight, age, etc). The applications then relate particular metadata values (e.g. ‘interests’, ‘age’ and ‘location’) to create connections between records.
You’ve probably guessed by now - the Semantic Web can do this, but better! For example, you would only need to store a single profile, which you could own/control. Social Networking applications could then be created by anyone, not just the few companies that store your profile secretly in their database.
-
And finally, Mash-ups. As users demand more access to data from companies, the forward-thinking amongst them are responding by providing API access to the data. We’ve then witnessed the emergence of ‘mash-ups’, that use multiple APIs to combine and relate data from distributed sources, into a single interface.
System Integration (of which this is an example) is yet another problem that the Semantic Web is looking to solve. The solution resembles the SOA vision that was/is dismissed by many. In theory, Semantic Web data sources can be very easily exposed and queried through a consistent interface (SPARQL, HTTP). As such, the Semantic Web actively promotes (in fact, relies on) users ‘mashing up’ data from multiple sources.
So there we have it - a limited “Semantic Web” appearing without the complex technologies that have been developed for it.
Will the trend continue? Can it, using existing technologies, or will developers eventually ‘find’ RDF, OWL and SPARQL? Should the appearance and popularity of these applications give Semantic Web proponents hope that the real thing is coming, or do they give Semantic Web opponents valid ammunition against unnecessary complexities?
I’d be interested in hearing your thoughts below.
It is very tempting to agree, and indeed I would do so, though I would like to add two observations. First: What can be reported on the agent issues (remember TBL vision?), and second: do all kind of data/metadata get the same attention, that is, who is aware of e.g. activities as W3C Multimedia Semantics Incubator Group .
Cheers,
Michael
Ummm... if I write a foreward to a book describing myself, why I wrote the book, who I think influenced my thoughts in writing the book, and thanking my friends and wife by name, is that a semantic web technology?
If you separate the vision from the technology, what you have is common practice. See Integrated Open Bibliographic Hypermedia.
Escape the trap of the closed system of definitions that is the Web. A world of possibilities will open up.
It seems worth noting that it's been about 7 years since Tim B-L announced the Semantic Web vision the the core RDF technologies were made available. In the 7 years after he announced the Web vision and made the core technologies available, they literally took over the world, whether the world was ready or not.
The fact remains that Google et al do just fine without using RDF, OWL, SPARQL, etc. What evidence is there that they would do better if they did?
one thing we have to think about is that most of the developers of the current semanticweb-look-alike applications don't come out of a software-engineering environment. so they dont think about concepts and graph-theory. they want to implement their ideas and the easiest way to do this, is to use existing technologies which are easier to handle.
we have to take rdf&co 'down to earth' so that it become more attractive for the masses - especially for novices.
everyone who is interested in developing for the internet knows about php, perl or mysql. he can query the database or creating his own database-shema for his project. but why does he know about this? and why doesn't he think like "hey, maybe i can use a rdf-database or maybe i can create an ontology"?
its because no one shows him, how he can use these technologies in an easy and naturally way.
we need a lot of easy examples, and scenarios which show how to use semanticweb-technologies - like all the examples which exist for the other technologies. then rdfs,rdf,owl etc. will come to their mind and many people will use it. and they wouldnt treat it as an exotic technology which only exists in universities, w3c-specifications and workgroups.
so i think that the trend will continue unless we change something at this level.
cheers.
daniel
Hello,
Everything ok, but linking money, what about it ?
Regards/Filipe
Maybe there is something vital we should learn from the web2 community. Web2 is about cooperation between human being, and semantic web so far is more about cooperation between machines. When we come to thinking about cooperation between human being, there are something different we need to consider. That is what the great success of web2 should teach us. Of course, semantic web can improve the current web2, but web2 definitely has something that we miss in semantic web.
I think that the web has grown in much the same way as a city - a city that has very rapidly become a metropolis. Many of the common problems in a large city like traffic jams, crime and bad neighborhoods, and building code violations, have obvious corelations on the web. Those conditions exist because it is impossible to build the perfect metropolis and then fill it with residents. Instead, the city is cobbled together and expanded as the need (and funding) arises. Without getting into the details of the analogy, I think it's highly likely that the web will continue to evolve in a manner that will stretch and tweak current systems (which are good enough to please most people and make businesses tons of money) and which precludes any utopian transformations.
The history of technology has long been filled with derelicts, grim and foreboding relics that attest to the dangers of creating technology that is too complex or difficult to maintain. The "formal" semantic web seems very much to be one such example, as much as I hate to say it. RDF (and the hideousness of using namespace URI as identifiers) strikes me as being a lot like Miranda or Haskell - capable of eliciting deep understandings about the nature of programming, but in general found only in academic realms or buried deep in highly complex applications that are fundamentally closed to the outside web.
Personally, I like Haskell - and I think there are pieces of it (such as the notion of monads) that are beginning to manifest in the world of both web and commercial development. However, its unreasonable to expect that Haskell will suddenly take off in the public imagination largely because it is too cerebral and complex a language for most developers to readily adopt. I believe that RDF and the Semantic Web in general falls very clearly into this category itself.
It used to be that to be in computer science you needed to have a solid grasp of finite mathematics - group theory, lie algebras, graph theory, combinatorics, etc. By and large I've found that those people who have familiarity with these areas (and who have a fairly sizeable grounding in computer programming in general) are readily able to handle the abstractions necessary for SW programming, and those that don't, aren't. This bespeaks a complexity level that is fairly unsurmountable for adoption, which is one of the reasons that I suspect that the gulf between the informal vs. formal semantic web will continue to widen over time, not narrow.
I agree with Dan and Kurt. While the RDF triple (subject, predicate, object) is easy to grasp, the learning curve for the SW stack is just way too high for your average geek to grapple with. We need tools and technologies that hide a lot of the ugliness of RDF, OWL, etc. before rapid adoption of these (underlying) technologies takes place. I want a set of ontologies that have just enough complexity to allow me to build pratical applications, and I don't want to screw around with APIs like Jena and Redland. They're too cryptic. Most application developers don't want to think in terms of RDF graphs, statements, etc. They want to work with tangibles like Person, friendsWith, worksFor. To hide RDF we need to take a cue from (I'll say it) Google -- the GWT abstracts AJAX. I like what the ActiveRDF guys have done to take advantage of the language features of Ruby to make RDF easier to work with. If I'm not mistaken, they actually built on top of Redland. That kind of thinking is definitely a step in the right direction. Building SW apps needs to be easy and the incentive needs to be obvious.
Ton's of GREAT comments from folks who know WAY MORE about this stuff that I do, inlcluding (and especially) yourself (refering to Dan for those wondering.)
If I could throw in one simple thing...
If I can't view source, copy/paste/go, then I'm not going to use the technology. I can always figure it out later, and in fact, thats how many of us, if not all, learned how to write HTML/CSS/etc... to then become "experts" in these, and other related technologies.
As other have noted, the barriers of entry to RDF and related technologies is simply to high to reach ANY level of critical mass.
One other note: In his now infamous MySQL keynote from last year, Adam Bosworth described the process of Google's search utilities as putting PhD's in tanks and having them drive through brick walls.
The job of the semantic web technologists is not to develop new ways of "connecting the dots" it's to find way's of hide the details and build from things we are all already used to using. Like a text input box and a "Search" button.
A FANTASTIC example of taking things to the next level without retraining the end user how to "think"...
Google Suggest > http://www.google.com/webhp?complete=1&hl=en
Yes, you're right, the world is heading towards deploying semantic web
principles, but that doesn't mean that this will eventually lead them
to RDF, OWL and company. These are brilliant technologies, but they
are evolutionary dead ends which are based on assumptions about how
the Internet works which aren't valid any longer.
Distilling meaning from chaos is a noble goal. But meaning handed to
people on a plate is meaningless. You can't understand something
unless you've gone through the process of evaluating, structuring and
distilling that information. It's the process that leads to
understanding and meaning.
The semantic web would at best reduce the Web to something like an
ancient greek oracle. It will give you an answer, but it will still
take an enormous amount of work to figure out what that answer means.
The process is hard work, which is why so many people trying to
develop universal metadata and cataloging systems in the '90's
failed. So everyone threw up their hands and said people are too lazy
to do metadata (as was put brilliantly in Metacrap) but this was
before Wikipedia, before social networking, before tags. People will
do metadata and catalog, but not for everything. But if everyone did
at least a little metadata and cataloging and then pooled the results
you can end up with an impressive amount of metadata.
http://blog.tomevslin.com/2006/06/price_the_cost_.html
There is a single thread running through most of these dicsussions which the blogger gets: the cost of complexity.
OTOH, complexity is an output, not an input. To repeat: there is an inverse relationship among subjective views and objective implementations. As an object subsumes more operations (increases the semantic tensor) the topical/useful distance between it and the aggregate set of users grows exponentially. More scope = less reach.
Be less greedy and you become less needy. That is how networks grow. ;-)
The difference between Web 1.0 and Semantic Web is that TimBL has implemented and released (free) Web browser AND Web server for the original web. For Semantic web there is a Vision and Specs released. Geeks like code, people like free utilities, and big corporations like (their own) specs standardized.
Google IS using Semantic technology. In fact, almost all money they make is from the Ads technology based on the large acquisition Google made: Applied Semantics. http://www.appliedsemantics.com/
The "trick" is that Semantic technologies work well in controlled system (like Google Ads), with optimized platform, rather than XML representation of RDF. The idea of RDF is great. The XML implementation is not.
By the way, relational databases are also not good "fit" with RDF either. But since there IS a demand for RDF (government, financial, pharmaceutical industry), Oracle does support RDF in "native" (spatial/network DB) way.
So Web 3.0 will make some smart usage of semantic data, and it will be an evolution of what we already have, where the specs and marketing will follow simple and smart hacks. Like Google maps prompted Ajax for Web 2.0. Just don't expect it to be called Semantic Web, and it is likely to be more messaging/P2P than client/server.
I think it takes people time to find the usefulness of the new standards. CSS was around a long time before people actually figured out why it was created and started using it mainstream. Give it another couple of years and everyone will be using the new standards.
WWW/Internet is about cooperation of people. So the key is how to involve normal common people in the process of cooperation. If you can have a way to leverage cooperation, it will easily scale up and produce a miracle. I feel that is why web2 succeed (also the reason why TimBL succeeded at web1). Although semantic web promise a great future, it miss a big point. The world in which people can cooperate more freely will be a future more promising than the world promised by semantic web. Finally the true nature of computer science emerges, as a bridge between physical science and human science/arts.
I agree with what Brad said. Yes, it is a human process. The key is to make things easy for people to understand and use so they can easily cooperate with each other. The complicated science only works behind the scene.
There are definitely some lessons that semantic web community can learn from web2.
Hello,
My project is about linking marked money : to datevalue it and add new Economy 4G3W.
Regards/Filipe
Your commentary is spot on! The Semantic / Data Web is here. Web 2.0's eventual limitations will set the stage of the final unravelling. Web 2.0 users are already beginning to realize that current Web 2.0 offering are new frontiers for lock-in (you data ends up in a web accessible silo that's incapable of providing querying expected data access flexibility).
The real key to the Semantic / Data Web bootstrap is the emergence of shared ontologies (FOAF and SIOC are great examples). SIOC and FOAF will deliver the same community bootstrap that RSS provided to Web 2.0 via the Blogosphere. Just watch!
To Michael Champion: Google et al are not doing fine for the reasons you give (they are doing fine in a different way as mentioned at the end of this comment).
There is much more to the Web that what Google currently offers. XQuery and XPath don't cut it either. When you juxtapose GData and SPARQL I hope my commentary will become clearer. Please do not assume that the Semantic Web implies user / consumer level knowledge of OWL, RDFS, RDF etc.. That's for technology providers to handle. You can't tell me that you genuinely believe that users and developers aren't seeking easy ways to query Blogs, Wiki, Feed Aggregation, Bookmark, and other Data Spaces without the current limitation of limited search services provided by Web 2.0 services.
As I stated a while back in my podcast interview with Jon Udell, SPARQL is going to help you locate relevant data across the Data Web which, if well-formed, will benefit from additional data extraction via XQuery/XPath. These things work together ultimately (including Web 2.0 itself which is itself the ultimate usecase creator for the Semantic Web).
BTW - The existence of Google Co-op is a tell tale signal that they know what they are offering isn't good enough since this is a clever effort to accelerate user driven semantic content creation and annotation behind the scenes.
Skip all comments and go straight to reading Kingsley Idehen. Any Valley insider knows that Radar Networks will change everything. I hear that Google has an offer to acquire the company for a Facebook
valuation. This is the de facto SW platform.
You think we "skipped" the SW technology ?
So there we have it - a limited "Semantic Web" appearing without the complex technologies that have been developed for it.
Will the trend continue? Can it, using existing technologies,(...) ?
So what is RSS (be it 1.0 or not) other than what is done with RDF in "higher" SW applications ?
I think one key issue with the adoption of W3C's Semantic Web is that programmers like programming and not philosophing about potentially arising AI issues.
We probably can compare this to other complex tasks like e.g. Geomapping. There were few useful applications of it until Google launched their mapping service.
maybe this is an option for many semantic applications: Have a web service programmed to do the hard (reasoning and infering that is) stuff and have it easily deployable via standard practice interfaces for all the others.
It's a bit what we are trying to do with our SemaWorx Project (http://www.semaworx.eu/): hiding the strong stuff and processing details from the user, while enabling both the average web user and programmers to profit from the result.
Any comments on this approach will be highly appreciated !!
Torrie Wilson in a thong
http://torriewilson.cn.to
Semantic web will emerge in 2 fronts:
- the acceptance of revolutionary server side technologies such as RDF, OWL, and SPARQL. You have to build RDF and OWL knowledgebase from scratch because migrating the existing Data Web Databases into RDF and OWL knowledgebase is not feasible. The RDF/OWL knowledge acquisition/maintenance and SPARQL knowledge use are already in place - once these processes become universal due to better tools/utilities available, the revolution of the semantic web's server side technologies will happen overnight. Usability of these server technologies is not an issue because it is hidden from front end developers and end users. The problem right now is that the hard-core back end architectes/designers and developers are still too lazy to pick up concepts and technologies that RDF/OWL/SPARQL currently deliver. We need some fine evangelists to do the groundwork here to push it forward.
- the emergence of evolutionary client side technonologies that handle request/response from the revolutionary server side technologies. It must be evolutionary because front end developers cannot embrace the revolutionary server side technologies on one day. The revolutionary server side technologies will give birth to new client side technonologies whcih usability becomes critical and crucial to the success of the Semantic Web. We do not see any of these client technologies for the semantic web today. It must happen in the evolutinary fashion. However, first, hard core backend architects/designers/developers must embrace RDF/OWL/SPARQL first before they can create new client technologies for the semantic web.