Henry Story: "Luckily the Semantic Web at its core is really simple, which is exactly what makes it bullet proof"
01.02.2007
Henry Story is a Staff Engineer at Sun Microsystems where he does Semantic Web research, development and evangelism. He is working on a number of open source technologies and demo applications as they help make the case for powerful ways of helping engineers use the semantic web to their advantage. Henry gave an interview to the Semantic Web School on a variety of questions around the semantic web.
SWS: "Henry, what is the exciting thing about the Semantic Web?"
Henry: The Semantic Web brings together some of the most interesting fields in computer science of which: the web, data bases, object oriented programing and reasoning. Having studied logic, philosophy, and computer science, spent a lot of time programming in Java since its earliest days, and worked for a large web company (AltaVista) I can see how the Semantic Web is building something completely new and with huge potential on a world wide/web wide scale. The more I discover about it the more it looks like it is a distillation of all the knowledge and research in all those areas over the past 30 years.
It can only bring these fields together by being extremely simple and clean. And this is what I like most about it: when you peer past the proofs available on the w3c (and mathematicians can spend a page proving that a circle is round btw), the concepts are really very simple, building upon what has to be the hippest thought around: everything is connected.
SWS: "Sometimes a 'semantic evangelist' like you might get tired of explaining what the semantic web is all about. So, how do you address that problem?"
Henry: I try to blog about everything I discover. Then I can refer people to things I have already written and avoid too much repetition :-)
At Sun my boss Hal Stern, who was CTO of software kept asking me for 10 minute presentations on the Semantic Web, so I have also gotten to be very good at explaining the core of the Semantic Web very quickly. To do this I have to simplify the ideas down to the core. And luckily the Semantic Web at its core is really simple, which is exactly what makes it bullet proof.
But now I know about Semantic Web School I may, if I get really tired or have too many people to convince, just send people your way ;-)
SWS: "What are your favourite arguments for the use of semantic technologies, especially when you have little time to explain?"
Henry: "Well it all depends on where the person I am trying to convince is
coming from.
- A business person may be more interested in understanding why
connected information is more valuable than isolated information. For
them I have an extension of Metcalf's law as applied to the
Semantic Web. The value of connected information grows
exponentially with the number of connections it has to other
information. Allowing information to be easily connected is the best
way to grow the value of the information, and so the visibility of
the goods the information is speaking of. The Semantic Web is about
making information easily connectable. - People with more of a database or web 2.0 background will find
SPARQL the most interesting technology. As businesses open up
relevant parts of their databases to querying, small services
will be able to spring up and mash this data up with other
information to cover up each of the many business opportunities in
the long tail. And there are many. SPARQL can return Javascript
result sets for easy parsing by browser agents. The browser agent no
longer needs to learn a million different query languages to query
the web of data, the database manager no longer has to develop a
query language for those users each time he wishes to put a service
online. Even schemas (Ontologies in semweb talk) can be developed in a distributed fashion and the collective experience be reused easily,
as with OO programming. - A good demo after a quick introduction helps a lot. I am working on a Semantic Address Book which can demonstrate immediately the value of the the millions of rdf foaf files out there in one Drag And Drop gesture. The D2RQ server I place online can allow people to compose their own SPARQL queries and learn by playing. The server is not on a fault tolerant server, and so can go down without my noticing. A more reliable way to play with SPARQL queries is to use a nice UI called twinkle.
SWS: "To build data-models by using the W3C Semantic Web Stack, the introduction of an 'RDF-Bus' might seem to be a bit unusual to some DB-specialists. So what are your arguments for doing so?"
Henry: The RDF-Bus is in fact very easy to explain to data-base specialists
when one presents them with a tool like D2RQ.
A relational database is so called because it is full of relations. Each column collects a set of relations, one for each row. The relation expressed in each row is the relation between the value of the primary key column and the value of the column in question, the relation type being defined by the name of the column. So in the following table
|id | sex | age |
---------------------
|1 | Male | 29 |
---------------------
|2 | Female | 27 |
the "age" column collects all the "age" relations. The second row states that the object identified with id 2 has age 27.
The semantic web uses URIs both for the primary key and the column names. Why does it do this? Because we need to be able to identify equivalent relations or objects around the world as being the same, so we need Universal Identifiers! Wanting to link up information around the world, URIs are the obvious solution. The rest pretty much follows from that...
SWS: "On what projects in the field of the semantic web are you working at the moment?"
Henry: I am currently working on a Semantic Address Book called Beatnik (BAB) which should have similar functionality as the Apple Address Book but be able to read any of the well over 20 million foaf files in the world, and write one too. This should help demonstrate how desktop apps can use the Semantic Web information and demonstrate a very powerful concept of Universal Drag and Drop at the same time. Hopefully we will be able to get everyone a URI at Sun so that data can refer to people using these uris, and so that by simply dragging and dropping those uris we will easily be able to add people to AddressBooks, mailing lists, chat conversations, etc....
To help me along in such an application I have written a java to rdf mapper called so(m)mer, an rdf equivalent of Hibernate, to help with the task of writing such an application. By adding simple @rdf annotations to java I can write code using simple POJOs without having to deal with background database complexities.
SWS: "Social tagging is quite a simple way to enrich information semantically. Do you work on concepts which will help to process tags in a corporate semantic web @ Sun?"
Henry: We have build some tagging engines in Sun, and at a recent meeting we had found that this could be extended in a very interesting way with wikis to generate useful vocabularies which could be the base for product ontologies.
SWS: "What are your recommendations to other globally acting companies concerning the use of semweb-technologies - when should they start thinking about a 'semantic shift'?"
Henry: My recommendation is: Learn about it now. Start some small projects to get a couple of people to understand the technology.
One very simple way to start is to just export data using a SPARQL endpoint, believing in Metcalf's law.
Think about the following: every product the company produces should have a permalink, a url and a semantic description at that url containing the technical product information, and more (such as links to components), so that other pages can start linking to them. With such links clients of the company will start finding it easier and easier to find product information, people who are speaking about that product, using it, etc...
But you will need people who understand the technology to understand how to best apply it to your company's needs. And it is really good to see that more such people are going to be available as Semantic Web Schools are built up around the world.
Don't panic: There are further links to Henry Story
