Peering Editor: Charles Petrie • [email protected] Embracing “Web 3.0”

Ora Lassila • Nokia Research Center • Rensselaer Polytechnic Institute

n an article published in The Times representation (KR), which is a subfield of artifi- this past November, reporter John Markoff stat- cial intelligence (AI) concerned with constructing I ed that “commercial interest in Web 3.0 — or and maintaining (potentially complex) models of the ‘,’ for the idea of adding mean- the world that enable reasoning about themselves ing — is only now emerging.”1 This characteriza- and their associated information. As such, we can tion caused great confusion with respect to the understand the Semantic Web through the lessons relationships between the Semantic Web and the learned from the Web’s development and adoption, Web itself, as well as between the Semantic Web as well as (perhaps somewhat painfully) from the and some aspects of the so-called Web 2.0. Some deployment of AI . wanted to reject the term “Web 3.0” as too On the Web, we’ve seen the emergence of some business-oriented; others felt that the vision in the completely new business models that do indeed article was only part of the larger Semantic Web work, despite initially seeming infeasible. These vision, and still others felt that, whatever it was include the models introduced or perfected by called, the Semantic Web’s arrival in the Business Netscape (creating a community by giving stuff section of The New York Times reflected an impor- away), Amazon and eBay (marketplaces), and tant coming of age. Yahoo! and Google (advertising-supported sites). With the Resource Description Framework Sharing data (or content, as it’s often called when (RDF) and (OWL) — the discussing the Web) has unexpected and serendip- languages that power the Semantic Web — becom- itous outcomes — once you make something avail- ing standards and new technologies reaching able, you have no idea how some people will use maturity for embedding semantics in existing Web it. The long-tail phenomenon — for example, pages and querying RDF knowledge stores, some- aggregate sales of low-selling items, such as spe- thing exciting is clearly happening in this area. cialized books, surpassing the total number of best- sellers sold — defies traditional thinking about Semantic Web Background business models, but it’s important to the new Web- With more than 10 years’ work on the Semantic based economy. Web sites don’t really exist in iso- Web’s foundations and more than five years since lation — linking is what makes search engines work the phrase became popular, it’s an opportune and gives the “blogosphere” its power. moment to look at the field’s current state and From the euphoria surrounding AI in the 1980s future opportunities. From a humble beginning as through the hangover of the “AI winter” in the a methodology for machine-interpretable meta- 1990s, we’ve learned what doesn’t work: you can’t data and through a “world-embracing” vision of a sell a stand-alone “AI application.” These tech- new era of software (often — erroneously, in our nologies make sense only when embedded within opinion — attributed as science fiction), the other systems. Tools are hard to sell and often fail Semantic Web has matured into a set of standards to make good business sense (they certainly don’t that support “open” data and a view of informa- make sense according to venture capitalists). Final- tion processing that emphasizes information rather ly, thinking of AI itself, we observe that reasoning than processing. engines are a means to an end, rather than the end From one viewpoint, the Semantic Web is the itself; how you use them is more important than symbiosis of Web technologies and knowledge the mere fact that you use them.

90 Published by the IEEE Computer Society 1089-7801/07/$25.00 © 2007 IEEE IEEE COMPUTING Embracing “Web 3.0”

More recently, many people have siderable effort in revisiting and store data in a flexible schema so you can become excited about Web 2.0. Al- extending the functionality in the store additional types of information that though we abhor both the term and OWL standard, which is now emerging you might have been unaware of when you the use of version numbers, we see as OWL 1.1. originally designed the schema. The second that the movement is rife with inter- Given that much of the current is that it helps you to create Web-like rela- esting phenomena. Web 2.0 is mostly work was presented at academic con- tionships between data, which is not easily a social revolution in the use of Web ferences, that new journals have sprung done in a typical relational . technologies, a paradigm shift from up relating to semantic technologies, the Web as a publishing medium to a and that much of the language design As RDF acceptance has grown, the medium of interaction and participa- happened in academic labs and corpo- need has become clear for a standard tion. From the Semantic Web view- rate research centers, some have under- query language to be for RDF what point, however, the most interesting standably assumed that the Semantic SQL is for relational data. The SPARQL technical aspects are Web is primarily a research vision that’s Protocol and RDF Query Language not yet ready for prime time. However, (SPARQL),3 now under standardization • (or “tagging”) pro- we’re starting to see considerable devel- at the W3C, is designed to be that lan- vide an organic, community- opment within the applications space guage. As Nova Spivack, CEO of Web driven means of creating structure and, as the “Web 3.0” article revealed, startup Radar Networks (www.radar and classification vocabularies; they often succeed where tradi- tional mechanisms for defining Web 2.0 is mostly a social revolution in the ontologies have failed or at least proven cumbersome. use of Web technologies, a paradigm shift • — the use of HTML markup to decode structured data from the Web as a publishing medium. (with the underlying thinking that human-readable representation now comes free) — are a step toward this work is emerging in an important networks.com), put it in a February “semantic data.” Although not in and exciting way. 2007 blog, “There is a huge amount of Semantic Web formats, microfor- interest in SPARQL at the moment, and matted data is easy to transform Web 3.0 there are already a growing number of into something like RDF or OWL for Although Semantic Web proponents SPARQL endpoints popping up around Semantic Web agents to process. have long seen evidence of growing the Web. These new SPARQL endpoints W3C is working on new approach- interest, the ’s success has are to data what Web sites were to es, such as Gleaning Resource become far more evident in the past documents.” Descriptions from Dialects of Lan- few months. This is largely because of Numerous players of various sizes guages (GRDDL) and RDFa, to stan- the maturing of the RDF languages and are now focusing in different areas of dardize the linking of structured the technologies that support them. the Semantic Web space. UK-based data with instructions on how to Oracle’s July 2005 release of RDF Garlik (www.garlik.com), for example, transform or embed data into exist- support in its Spatial 10.2g database uses Semantic Web technologies for ing Web resources. product provided the legitimacy that the “control of personal data in the some felt the language lacked. As peo- digital world.” Specifically, the compa- Since the 2004 completion of the ple experimented with RDF , ny is working to let users discover RDF and OWL standards, we’ve seen a they found significant advantages over what’s known about them on the Web lot of experimentation (and confusion) traditional structured databases in to see what the aggregation of this regarding the right representation lan- many cases, especially with respect to information (exposed via an RDF guage to use for any particular appli- embedding data on the Web. As store) reveals. Dave Beckett, an engi- cation. Not surprisingly, subsets and Microsoft put it in its December 2006 neer at Yahoo announced in Novem- extensions of these languages have Connected Services Framework 3.0 ber 2006 that the Yahoo Food site started to appear — most notably, ver- Developer Guide:2 (http://food.yahoo.com) is being pow- sions of RDF(S) that borrow a small ered by OWL and RDF, as well as sev- number of features from OWL (though There are two main benefits offered by a eral other technologies. Teranode remaining simpler than “OWL Lite”). profile store that has been created by using (www.teranode.com), among others, is Other developers have invested con- RDF. The first is that RDF enables you to exploring the use of Semantic Web

MAY • JUNE 2007 91 Peering

Browser HTTP Ontology with SPARQL Back Forward Stop RefreshHome Print Mail @ AI application Address: > go Dynamic HTTP RDF often beyond your ownership or control content Dynamic RDF triple (and potentially hostile), requires funda- engine Code content RDF HTML and store mentally new approaches to ensuring engine triple reasoner interoperability. No longer can we Internet Zone store expect a priori standardization of every Figure 1. Sample “fractal” architecture of Semantic Web applications. Dynamic pair-wise interaction between all possi- content engines, backed by RDF triple stores, act as both producers and ble systems we anticipate interacting consumers of “semantic” data. Data exchange can be facilitated using, for with; indeed, we can’t even anticipate example, SPARQL. all future scenarios. Operating in such an open-ended world requires mechanisms for limiting technologies for scientific data integra- an RDF triple store: a component that the decision-making scope. For exam- tion, particularly in the biology sector. allows, among other things, the integra- ple, when seeking a particular kind of Joost (www.joost.com), the new Inter- tion of heterogeneous data sources and new service to use, you’ll want to limit net TV platform that made big news in repositories. SPARQL’s emergence as the the set of candidates to something that’s February in announcing a partnership standard query language for RDF lets contextually relevant (such as those with Viacom, uses RDF extensively. In many data stores expose themselves as that are offered in your current loca- fact, Joost announced recently that it SPARQL endpoints, thus enabling flex- tion). Similarly, traditional access-con- will provide its open source RDF back- ible data exchange among systems. It is trol mechanisms might not scale to end technology to the Apache Founda- leading the way toward Web applica- situations in which we have an open- tion (www.apache.org), making it tions that exhibit a kind of “fractal” ended set of systems and users: we need much more widely available for use by structure, with patterns in which one new decision-making mechanisms to Web developers. component uses another as a data enforce more flexible policies. From the It’s interesting to note how little of source (via SPARQL, for example) and representational viewpoint, Semantic this effort focuses on what was once acts as a data source to yet another Web technologies offer the possibility thought to be the Semantic Web’s component (see Figure 1). Such archi- of implementing these kinds of techno- major business sector: the integration tectures open up new possibilities for logical frameworks and platforms. We of enterprise data assets via ontologies. the original vision of Web services and claim that context-awareness and poli- It isn’t that such work isn't going on — loosely coupled distributed systems. cy-awareness are complementary rather Oracle, IBM, and several startups are Essentially, we can view Semantic than separate mechanisms — think of all providing important capabilities in Web technology as a novel approach to policies (and their enforcement) as a that area — but embedding RDF and interoperability: application developers particular kind of context. OWL on the Web, via the all-important can defer to the runtime accessible In the longer term, given that URI mechanism, is a key part of the semantics of a dialogue between two Semantic Web technologies are matur- emerging excitement over Semantic information systems even after the sys- ing as a means of describing things, we Web technologies. Whereas the re- tems have been deployed. By using rea- could use their representational power search community is widely exploiting soning mechanisms to access implied to describe things in the real world. One the AI technologies that motivate, in information within conversations of view is that the physical objects will particular, the OWL DL sublanguage, the explicit statements, and by enabling become Web-accessible in that we will languages’ more “Webby” features — systems to dynamically add capabilities be able to represent them via . sometimes referred to as the “dark by acquiring new ontologies and data to Just like applying semantic technolo- side” of the Semantic Web4 — are pow- reason over, the Semantic Web lets us gies to problems of interoperability in ering the Web 3.0 technology space. build future-proof systems that have a ubiquitous computing environments, chance of “doing the right thing” even describing physical things will expand Beyond Web 3.0 in unexpected situations. This approach our scope beyond the current Web. This How do we see the future of the Seman- is particularly amenable to scenarios is not unlike when some argue that Web tic Web and, most importantly, the and situations in which interoperability services merely exploit mechanisms and application of Semantic Web technolo- is critical — for example, the ubiquitous technologies developed for the Web, but gies for “mainstream” IT problems and computing vision of environments with really have nothing to do with it. systems? With Web 3.0, these technolo- pervasive embedded computation. To Semantic Web efforts provide an ap- gies are finding fertile ground in multi- connect, say, your handheld device to a proach to constructing flexible, intelli- tiered Web applications in which the dynamically changing set of dozens, if gent information systems; some are middle tier can be implemented using not hundreds, of other systems that are Web-based applications, but we’re cer-

92 www.computer.org/internet/ IEEE INTERNET COMPUTING Embracing “Web 3.0”

tainly not limited to those. At the same “SPARQL Query Language for RDF,” W3C Lassila's work in the 1990s on machine- time, the application scope for Web working draft, 26 Mar. 2007; www.w3.org/ interpretable metadata for Web resources technologies is expanding elsewhere. TR/rdf--query/. contributed to the Semantic Web’s launch. For example, W3C has started an initia- 4. J. Hendler, “The Dark Side of the Semantic He is an elected member of the W3C Advi- tive dubbed the Ubiquitous Web, ack- Web,” IEEE Intelligent Systems, vol. 22, no. sory Board. Contact him at www.lassila.org. nowledging the benefits of expanding 1, 2007, pp. 2–4. the Web’s reach beyond our desktop 5. T. Berners-Lee, J. Hendler, and O. Lassila, “The James Hendler is the Tetherless World Chair at and laptop computers to other types of Semantic Web,” Scientific Am., May 2001. Rensselaer Polytechnic Institute, where he devices and situations. The synergies directs the Future of Information project and between ubiquity and semantics are an is a research fellow at the Nokia has appointments in the exciting area in which we expect to see Research Center in Cambridge, Massachu- and cognitive science departments. He also significant future work. setts, and is completing a PhD in computer serves as the associate director of the from the Helsinki University of Science Research Initiative, headquartered at Technology. Previously, he’s been a software MIT, and is a member of the W3C’s Semantic bout six years ago, we outlined a engineer, an entrepreneur, a venture capi- Web Coordination Group. Hendler is editor in A vision for the Semantic Web,5 talist, and has done research in knowledge chief of IEEE Intelligent Systems and serves including a view in which data des- representation and its application to produc- on the board of reviewing editors for Science. cribed in a machine-interpretable way, tion scheduling and logistics planning. Contact him at www.cs.rpi.edu/~hendler. coupled with a means for defining vocabularies and ontologies, would lead to a revolution in new Web appli- Advertising Personnel cations. In one of the article’s asides, Advertiser | Product Index Marion Delaney | IEEE Media, Advertising Director we reflected that we couldn’t really Phone: +1 415 863 4717 | Email: predict what the Semantic Web’s “killer May | June 2007 [email protected] application” would be. Rather, we claimed, “the abilities of the Semantic Advertiser Page number Marian Anderson | Advertising Coordinator Web are too general to be thought Phone: +1 714 821 8380 | Fax: +1 714 821 4010 about in terms of solving one key prob- Classified Advertising 15 Email: [email protected] lem or creating one essential gizmo. It Sandy Brown | IEEE Computer Society | will have uses we haven’t dreamed of.” Business Development Manager *Boldface denotes advertisements in this issue From enterprise data integration to the Phone: +1 714 821 8380 | Fax: +1 714 821 4010 coming generation of Web TV, the cur- Email: [email protected] rent variety of Semantic-Web-powered applications make it clear that this was Advertising Sales Will Hamilton [email protected] Representatives Phone: +1 269 381 2156 an understatement. Fax: +1 269 381 2556 Northwest (product) Although many aspects of the Mid Atlantic (product/recruitment) Email: [email protected] Peter D. Scott Dawn Becker Joe DiNardo Phone: +1 415 421-7950 Semantic Web are yet to be explored, Phone: +1 732 772 0160 Phone: +1 440 248 2456 Fax: +1 415 398-4156 Fax: +1 732 772 0164 Fax: +1 440 248 2594 Email: [email protected] and much research remains to be done, Email: [email protected] Email: [email protected] Southern CA (product) this technology is clearly transitioning New England (product) Southeast (recruitment) Marshall Rubin into a serious player in the modern Jody Estabrook Thomas M. Flynn Phone: +1 818 888 2407 Phone: +1 978 244 0192 Phone: +1 770 645 2944 Fax: +1 818 888 4907 Web universe. We might not like the Fax: +1 978 244 0103 Fax: +1 770 993 4423 Email: [email protected] Email: [email protected] Email: [email protected] term “Web 3.0,” but we enthusiastical- Northwest/Southern CA ly embrace the technologies it is bring- New England (recruitment) Southeast (product) (recruitment) John Restchack Bill Holland Tim Matteson ing to the field. Phone: +1 212 419 7578 Phone: +1 770 435 6549 Phone: +1 310 836 4064 Fax: +1 212 419 7589 Fax: +1 770 435 0243 Fax: +1 310 836 4067 Email: [email protected] Email: [email protected] Email: [email protected] References Connecticut (product) Midwest/Southwest (recruitment) Japan 1. J. Markoff, “Entrepeneurs See a Web Guided Stan Greenfield Darcy Giovingo Tim Matteson Phone: +1 203 938 2418 Phone: +1 847 498-4520 Phone: +1 310 836 4064 by Commonsense,” The New York Times, Fax: +1 203 938 3211 Fax: +1 847 498-5911 Fax: +1 310 836 4067 Email: [email protected] Email: [email protected] Business, 12 Nov. 2006. Email: [email protected] Europe (product) 2. Connected Services Framework 3.0 Develop- Midwest (product) Southwest (product) Dave Jones Steve Loerch Hilary Turnbull ers Guide, Microsoft, 2006; http://msdn2. Phone: +1 708 442 5633 Phone: +1 847 498 4520 Phone: +44 1875 825700 Fax: +1 708 442 7620 Fax: +1 847 498 5911 Fax: +44 1875 825701 microsoft.com/en-us/library/aa303446.aspx. Email: [email protected] Email: Email: [email protected] 3. E. Prud’hommeaux and A. Seaborne,

MAY • JUNE 2007 93