SIGSEMIS: Prof. Amit Sheth' s Interview: Semantic Technology is here to stay

AIS SIGSEMIS (SIG on Semantic Web and Information Systems, http://www.sigsemis.org)
An Interview with Amit Sheth: The Information Systems Perspective on Semantic Web Research 
 

'Semantic technology is here to stay...'

Full interview available at: http://www.sigsemis.org
 

Miltiadis: Dear Prof Sheth, we are really honoured for your kindness to provide us this interview. I would like to start by asking you what do you think for the so-called Next Generation Web Research. Do we have to wait for much more time before we will have an evidence of real world applications?

 

Amit: Thanks for talking to me. There are a number of exciting "next generation" technologies on the horizon that will drastically change how we see the Web, internet, computing and communication in general. Use of semantics (as in semantic technology or in a more focused perspective taken by some in SW community) is only one of the important component technologies.  Here again, semantics is only being exploited by a majority of SW researchers for a relatively narrow purpose of automating simpler- or shall we say well defined- things, although some commercial semantic technologies are being used for addressing broader objectives.  In the larger scheme of things, more automated travel reservation or scheduling is not something humans care so much about-I mean it is interesting for pedagogy but given that there are so many subjective elements in making a travel arrangement, not something that would be used in the way the problem is formulated. What really matter is the real purpose of any technology-e.g., for businesses it is gains in productivity and competitiveness, for a common man it is often about luxury, relaxation or entertainment, and so on.  Here automation is only a part of the equation. 

 

As for your question on evidence of real world applications - for semantic technology (as also relevant to the vision of the SW), applications are here already-paid for and deployed! Because of business considerations, most of these are Enterprise applications rather than pan-Web applications.

 

Miltiadis: Dear Amit, you are one of the most active people in SW and the most interesting is your IS perspective. What is your opinion for the role of  SW in IS research and vise versa? What do you answer to all those people who claim that OK.SW? But where is it?

 

Amit: Semantics has long been recognized to be very important in IS, Databases, AI, Linguistics and many other fields.  From the IS/DB perspective, I remember talking about "So Far (Schematically) yet So Near (Semantically)" in 1992, but lots of smarter people have talked about semantics for some time.  More recently however, two things have happened-one positive, one potentially not so positive.  The positive thing is that we have now been able to engineer semantic technology that supports large scale semantic applications, and use large populated ontologies to provide semantic underpinning.  At the same time a questionable development is a rather overwhelming importance attached to "formal semantics".  Tom Gruber who brought limelight to the term ontology in early 1990s last year talked about limited value and success of formal ontologies (something he had worked on) and underscored the importance of semi-formal ontologies.  I have further explored this theme, along with real world observations, in my article in Data Engineering Bulletin in December 2003. Underlying to my albeit personal view is that I must have the expressiveness in representation to meet my application requirements, rather than starting at the other end -- with computability and computing concerns-for example of an inferencing technique- and then set out to determine expressiveness and my modeling capabilities.  Thankfully, query processing works for a broad variety of data and knowledge representation, and we have learned to implement that efficiently. This is a key relevance of IS/DB for semantic technology.  As for SW for IS, we get to look at challenging applications where multidisciplinary approaches are necessary and hence learn to leverage techniques from allied areas.  For example, to develop solutions for some of the complex problems in integration analytics that we deal with now, one needs to bring together automatic classification using machine learning, NLP techniques for document processing, and IS/DB for semi-structured and structured data management and query processing.

 

 

Amit P. Sheth 
Professor, Computer Science, University of Georgia

Director, Large Scale Distributed Information System Lab

                        

Sheth is an educator, researcher and entrepreneur.  He joined the UGA and started the LSDIS lab in 1994. For nine years before that, he served in R&D groups at Bellcore, Unisys, and Honeywell.  In August 1999, Sheth founded Taalee, Inc., a Venture Capital funded enterprise software and internet infrastructure startup based on the technology developed at the LSDIS lab.  He managed Taalee as its CEO until June 2001. Following Taalee's acquisition/merger, he serves as the CTO and co-founder of Semagix, Inc (formerly Voquette, Inc).  His research has led to two companies, several commercial products and many deployed applications. He has published over 150 papers and articles, given over 130 invited talks and colloquia including 19 keynotes, (co)-organized/chaired ten conferences/workshops, served on 90 program committees, etc.