Jennifer Zaino
SemanticWeb.com Contributor

Auction giant eBay is taking to the semantic web for helping to document its systems as part of its grid computing initiative in its data center. Semanticweb.com recently discussed why and how with Neel Sundaresan, Director of eBay Research Labs.

Semanticweb.com: What is the main purpose of using semantic technologies at eBay?

Sundaresan: The first and the main purpose of using semantic technologies is documentation. We know that a well-documented system is a well-maintained system. As we have entities like operating systems, databases, servers, storage systems, and instances of these, we want to be able to describe two things -- what are their properties and how these entities are related to each other. When you want to create new instances you want to know conceptually and structurally where they fit. Hence, a well-defined object hierarchy becomes important.

More From Jupitermedia

Military, Universities Team Up on Big CALO Project Radar's Twine Ties the Semantic Web Together A Snapshot of Semantic Web Trends Semantic Web as Competitive Advantage Smartening Up Your Links If you want to comment on these or any other articles you see on Intranet Journal, we'd like to hear from you in our IT Management Forum. Thanks for reading. - Tom Dunlap, Managing Editor.

When you want to modify or delete instances you further want to know how that affects entities they are related to. In a complex large-scale system, defining these properties, inheritances, and relationships help us better manage, troubleshoot and scale the systems. Semantic technologies provide the right fit.

RDF provides the underlying framework for semantic description of our systems. OWL is a derivative of RDF with a stronger machine interpretability, which is important when you want software to interpret your rules to provide interfaces or take actions. We are both a consumer of these technologies and a key player in driving the Grid standards (OGF). [The OGF is the international community dedicated to accelerating grid computing adoption by providing an open forum for grid innovation and developing open standards for grid software interoperability; standards such as RDF can be used to extend current grids into semantic grids in which information and services are given well-defined meaning.]

Any well-managed and documented system should be using some sort of semantic technology. [But] people don't always wait for standards to come around to help them solve their problems. That is why many of these tend to be one-off implementations. The vision of semantic web included RDF as a syntax and schema framework for a metadata description language for the web. RDF itself was preceded by other work for non-web environments. RDF is just a framework and is not the only framework. Some [organizations] went with RDF while others did not. Specific instances of RDF have to be created for individual domains.

For instance, almost a decade ago, while some of us were using it to describe the Web, other standards were being formed. OSD (Open Software Description) comes to mind -- it was a language for describing contents and dependencies of software packages. It was used to deliver different types of software to diverse client platforms.
Well-known companies used it. Whenever you have a large complex system -- such systems are not uncommon in this distributed and networked world--these technologies become critical. As creations like literate programming helped software engineering, I see these technologies help in creating literate large-scale systems management.

Semanticweb.com: What are some of the considerations eBay has had about using semantic technologies?

Sundaresan: Evaluating and picking the right set of platform implementations and toolset is key. What is the right serialized form to use? What is the right query language to use? What OWL implementation to use? There is Turtle, Sparql, OWL-DL and several others -- few in the open source domain and some commercial. Some of them are incomplete or do not scale to the extent we want to scale in terms of storing, retrieving, querying, and modifying. We want to be sure that we use the right interface system to produce usable and scalable visual aspects of the system.

Semanticweb.com: What has been the ramp-up in terms of developing IT staff expertise in these new technologies?

Sundaresan: The main purpose of these technologies is to build tools that are built on top of the technologies. Building the right tools with the right sets of interfaces will make the job of the IT staff significantly simpler. Instead of writing one-off scripts for provisioning or following adhoc mechanisms to manage these systems, if the staff uses these tools, they are much better off. You have a better cataloging system, a better way of managing, of identifying problem areas and fixing them. You get a holistic view of the entire system, and the task only gets easier from there.

Semanticweb.com: Does eBay see potential for these semantic technologies to be used in other capacities in its business?

Sundaresan: Wherever you have data, especially in large forms, you have a purpose for metadata, with which comes semantics. And eBay has a lot of data to deal with. Whether it is systems information as we discuss, user, session, and transaction information, query and search, catalog, product and item information -- the list goes on. Better description, management, and scaling of all of these is critical. The models, purpose, and implementations might vary from use-case to use-case. And not all metadata is clearly definable by a team of administrators.

But the fact remains that semantic technologies will take us well beyond data and syntax, and applying the right pragmatics in the right context is our vehicle for ultimate success.