7 questions to Kingsley Idehen: "By declaring its context, Linked Data can be made more easily reusable by others"
15.06.2010
Kingsley Idehen is CEO of OpenLink Software and probably one of the most profound experts on integration issues. In business for more than 20 years he has observed the evolution of the Semantic Web from the beginning till now. Andreas Blumauer took the opportunity and asked the Semantic Web pioneer 7 questions about Linked Data, its relevance for business and the technological paths ahead.
1. faster, easier, and cheaper
A.B.: From a market perspective Linked Data and Semantic Web is still at the very beginning, although (large) enterprises or public administrations have recently begun to start projects around that approach. How can Linked Data help to make companies more productive? Please, give us your elevator pitch!
K.I.: Linked Data makes the discovery of relevant information faster, easier, and cheaper, whether that discovery is made through a precision search or a serendipitous connection.
2. Semantic Web is part of an innovation continuum that includes 40+ years of DBMS research.
A.B.: On some occasions you were pointing out the void between domain experts, their "internal" ways to represent knowledge and the way this is done by traditional (database) technologies. Do you think that W3C's specifications around the Semantic Web can offer a solution for this dilemma?
K.I.: I don't think this is an issue of specifications any more. The challenges lie in the alignment of terminology, and in generating an appreciation for the history of database technology. The envisioned Semantic Web is part of an innovation continuum that includes 40+ years of DBMS research.
3.a. In a nutshell: There are many waves to come.
A.B.: Even those people from the Semantic Web community who were promoting pure Semantic Web (OWL, Reasoning, Semantic Web Services etc.) for years and years have recently started to choose a rather pragmatic way to spread the idea (URIs, HTTP, SKOS, ...). Do you think that the Linked Data Initiative can build upon a stable architecture or will it face more and more problems the bigger the "cloud" will grow?
K.I.: Linked Data stemmed from a realization that the story of the Semantic Web was being told back-to-front.
Now that Structured Linked Data has been firmly established on the Web, other pieces of the overall vision can be built on this pragmatic foundation. As the Linked Data cloud grows, clear value propositions for other aspects of the Semantic Web will emerge. There are many waves to come, in a nutshell.
3.b. Context is vital to the comprehension and effective exploitation of all data.
A.B.: Which kind of semantics is still missing to organize the "mountains of brittle data" we are creating?
K.I.: Injecting context into Linked Data is something I've talked about for a long time, as what I call "Data Spaces." Basically, context is vital to the comprehension and effective exploitation of all data. By declaring its context (i.e., its Data Space), Linked Data can be made more easily reusable by others -- even in different contexts, because they can account and allow for the original context.
The hype around Linked Data can sometimes convey the illusion that the Linked Open Data community shares the same context, universally -- or equally, that it shares no context at all. Eric's post validly seeks to keep this important issue a part of the general community conversation.
4. security, performance, and manageability
A.B.: OpenLink Software is a leading provider of middleware software. The concept of "virtualizing" databases on a logical layer sounds familiar to any Semantic Web supporter. But how can one be convinced of such an approach? Who has been thinking the last 20 years only about things like security, performance, putting data into one manageable silo, and again: security?
K.I.: Security, performance, and manageability have all been key concerns in OpenLink Software's product development since day one. That remains the case today.
It is important to recognize that opening data silos in the sense of proprietary formats, access mechanisms, and the like, does not require exposing all the data so "freed" to any and all prospective viewers. Permission to use the virtualization layer to read from or write to the ultimate data store may be granted or not, as dictated by the policies of the data owner(s), based on the user, their location, the software they're using, or a myriad of other factors.
With regards to security, this is also where Linked Data offers very powerful infrastructure. Today, it's possible to associate HTTP based Personal Names (WebIDs), their host Structured Profile Documents, with X.509 Certificates. Then via the aforementioned associations embrace and extend SSL such that following happens:
A standard X.509 Certificate is created with an embedded HTTP URI-based WebID. The Public Key components (modulus and exponent) of the Certificate are written to a structured profile (typically FOAF) document available at an HTTP URL (Address). Protected Resources on Data Servers are given HTTPS based URLs. When those Protected Resources are requested, the Data Servers use a standard SSL handshake to challenge the Client connections for proof of Identity. During this handshake, the Client proves possession of the Private Key associated with a Certificate. An HTTP lookup is then performed against the Profile Document URL contained in that Certificate, to check whether the Public Key components in the Profile match those of the client's authenticated Certificate. These steps put an end to password-based security while laying the foundation for more socially-aware data access policies.
5. SPARQL enables enterprises to bring Relational Data to life.
A.B.: In the world-wide Semantic Web Community, Virtuoso has a strong reputation for high-performance queries over large RDF graphs. Nevertheless, graph-based data models are still not wide spread in the Enterprise World. What´s the ultimate argument for an Enterprise Architect to use languages like SPARQL at least in addition to SQL?
K.I.: The idea of Concept-focused Data Access is currently sweeping through the enterprise. The real-world orientation of graph-model database technology (as applied in the realm of RDF) allows us to offer conceptual-model access to existing data sources. Combining SPARQL with SQL (at OpenLink, we call this "SPASQL") enables us to deliver critical value to the enterprise, using a standard and well understood query language. Put differently, SPARQL enables enterprises to bring Relational Data to life without making them lose the time-tested virtues of SQL.
6. Change sensitivity, or freshness, is a critical factor in measuring "Quality of Data".
A.B.: Just recently OpenLink Software released DBpedia Live. Who are the people behind this project, what are their goals and how will a "Real Time Semantic Web" change the whole game?
K.I.: DBpedia-Live (as well as the original "Static Edition" of DBpedia) is a collaboration between OpenLink Software, the University of Leipzig, and Freie Univerity Berlin.
Data is a perishable good. Change sensitivity, or freshness, is a critical factor in measuring "Quality of Data". Until now, static Linked Data Warehouses, both public and private, have dominated the Linked Data landscape. DBpedia-Live is meant to highlight change sensitivity and help bootstrap a more dynamic Web of Linked Data.
The tight coupling of DBpedia and Wikipedia enables DBpedia-Live to add a very powerful zeitgiest-oriented dimension to DBpedia. Basically, as events occur globally, edits to Wikipedia will effectively construct multi-dimensional interest clusters that pull in contributions from traditional media organizations, in a mutually beneficial manner.
7. Thinking in terms of a Web of Linked Data.
A.B.: Last question, Kingsley: How will the "Semantic Web" be called in 10 years? Will there still be a "Semantic Web"?
K.I.: (...smiling...) I only think in terms of a Web of Linked Data.
About Kingsley Idehen
Kingsley Idehen is Founder, President, and CEO of OpenLink Software, a privately held company founded in 1992. Born in Manchester, England in 1965, of Nigerian heritage, Idehen was raised in both Great Britain and Nigeria. In the mid 1990s, he relocated to the USA, where OpenLink is now based.
Key Links:
1. http://www.openlinksw.com/blog/~kidehen - blog
2. @kidehen -- twitter
3. http://virtuoso.openlinksw.com -- virtuoso
4. http://www.openlinksw.com -- OpenLink