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.