KM World Sessions: Is Semantic Technology Real?

This is another in a series of notes from the 2009 KM World. It is titled: Is Semantic Technology Real? It is moderated by Rob Gonzalez, Platform Product Manager, Endeca Technologies and include: Michael J. Cataldo, CEO, Cambridge Semantics, Daniela Barbosa, Business Development Manager, Dow Jones Client Solutions, Dow Jones & Company, Lorenzo Thione, Founder / Principal Program Manager, Powerset / Bing Microsoft, Inc.  Here is the session description.

“Semantic technology is all the rage, sometimes even dubbed “Web 3.0.” However, many people—especially those making technology decisions for enterprises—wonder whether semantic technology has meaningful applications in the enterprise. Based on hands-on experience working with semantic tools, this panel of experts will establish the boundaries between reality and hype and help you understand what enterprises can gain from semantic technology in the here and now.”

Cambridge Semantics provides semantic development tools.  Dow Jones Client Solutions helps clients with taxonomy solutions, as well as other information needs. Powerset was acquired by Microsoft. It is Web semantic search tool.  Rob said, in answer to the panel title question, that semantic search has been applied successfully inside the enterprise and on the Web.

Rob asked the panel if there any examples of financially successful Web applications of semantic technologies. Lorenzo said yes. This happens more when you use semantic technologyto improve an existing site or product on such applications as consumer search. It seems harder for new sites such as Wolfram Alpha.

Daniela said in publishing there have been successful examples such as the Huffington Post and some Web advertising that use semantic web but do not promote it. The sites that promote that they are semantic Web are less successful.  Best Buy is using semantic web with their whole catalog but it is not promoted as a semantic web site. Google bought a semantic search company and it formed the foundation for Adsense and Adwords.

Semantic web appls let you combine different data sets like comparing the odds that you will be killed by sharks vs. vending machines.  It turns out that Vending machines are more deadly.  Semantic search helps people better define what they want in search.  The definition of semantic technology has been fluid.  Rob said it is everything from natural language document processing, and sentiment analysis to geotagging so it is metadata on top of data.  It is finding new meaning in data. Michael said semantic technologies are based on emerging standards on ontologies, distributed query, and other related things.  Semantics helps you find concepts even when there are not exact word matches. For example, you can find out how many sub-prime mortgages are in your loan portfolio even if they not are labeled that way. You define the characteristics and then find them. Here is a paper on semantic technologies and ontologies. 

The panel moved to semantics in the enterprise. There have been a number of successful examples where you can use a computer to connect data in two different places. How do you know that different things are similar? This is the answer that semantic technologies can address. It is about understanding concepts. FAST was given as an example. Consulting companies are starting practices around implementing semantic technologies.  At Dow Jones they are seeing companies interested in making better use of existing data through semantic technologies..

Daniela Is hoping that semantic technologies will get implemented like wikis, start small and spread.  This will be better than starting big.  Michael said that ontologies can grow over time. 

Some one asked what is new about semantic technologies? Daniela said the open vs. closed concept.  You can find better connections now. Companies do not know they are already part way there so some stuff is not new.  Now you can get your siloed data connecting and get better access to data outside the enterprise. Semantic technologies are trying to find structure in data where it does not explicitly exist in the same way a person might.

Merck is using semantic technologies to automate the integration of research results from three different relational databases and several spread sheets so they can be uploaded into a common database. Prior to this people had to cut and paste content by hand into a template for the upload.

Semantic technologies support discovery in contrast to simple search when you know what you are looking for. There remains a lot to do to improve the interfaces of semantic technology tools.