Smarter Searches: Technology Merges Images, Data and Knowledge
University of Texas at Dallas News Center
Let's say you're looking for a one-bedroom apartment in
northeast Dallas that has a grocery store, Catholic church and health
club within walking distance, costs less than $800 a month and has both
high-speed Internet access and a low crime rate.
Finding that dream apartment could take awhile right now. But
technology under development by UT Dallas researchers aims to speed
such a search, enabling people to find the perfect place today and move
Is that soon enough for you?
Whether you've just arrived to start studies at a university or you're
expected to start a new job on the other side of the country on
Tuesday, there's a good chance that sooner or later you'd make use of
such technology. Which is why Latifur Khan and his team of graduate
students at UT Dallas are working to provide options that take such Web
searches to the next level.
The secret lies in merging evolving semantic Web technology with
geospatial information systems.
The semantic Web is a smarter Web in which online content will be
machine-understandable, enabling computers to process the meaning of
words and phrases - and the thoughts they express - rather than simply
searching for keywords and phrases. Combine that with maps, photos and
other visual information, and what would otherwise be a complex search
is greatly simplified.
"We initially developed tools that showed apartment listings on a map
in a manner that would simplify the process of searching for an
apartment," Dr. Khan said. "The challenges that confronted the user
grew out of the raw and unstructured nature of the documents. We had to
deal with postings of all kinds of data, some having a physical
address, some having links to Google/Yahoo maps and others having the
name of a locality that only existing residents might be familiar with."
Dr. Khan's goal has been to create software that extracts useful
information from this array of content using a standard reference, the
U.S. Gazetteer. His team's work includes what's called the
disambiguation of names, establishing whether "Paris," for example,
refers to a city in France, a town in Texas, a famous socialite or the
guy who made off with Helen.
"As we make progress with our tools," Dr. Khan said, "we will add new
features such as locations of grocery stores, churches, temples,
mosques, synagogues, hospitals and other buildings of interest. We are
also working toward showing crime statistics for the neighborhood. Our
goal is to integrate the crime reports provided by the various police
departments on their respective Web sites, update these reports
continuously and summarize them based on a time window that makes sense
The research has already received quite a bit of attention from other
experts in the field, according to Bhavani Thuraisingham, head of the
Jonsson School's Cyber Security Research Center.
"In addition to publishing papers in several premier journals and
presenting results at conferences such as the Association for Computing
Machinery's International Conference on Advances in Geographic
Information Systems with researchers from the University of Minnesota,
the team is also developing tools jointly with Raytheon and
demonstrating them at major government conferences," she said.
The team has already developed a system called DAGIS (Discovering
Annotated Geospatial Information Services), a semantic Web framework
for geospatial information that the team subsequently extended to
handle queries related to police blotter data. The team is now
developing algorithms to facilitate integration of additional
geospatial data. The team has also been presenting research results at
government technology exchange conferences, and recently Ph.D. student
Jeffrey Partyka discussed the work at IARPA's Knowledge Discovery and
Dissemination Symposium at the Oak Ridge National Laboratory.
"These presentations and demonstrations have given us lot of visibility
not only with the research community, but also with the government,
vendor and standards communities," Dr. Khan said. "Our former student
Ganesh Subbiah was highly sought after by geospatial companies and now
works for ESRI, a leading geospatial information systems company. We're
also a member of the Open Geospatial Consortium and have given
presentations at the consortium's meetings."
"They bring strong geographic information systems expertise, which,
combined with our information management expertise, make us a powerful
team," he said.
The School of Economic, Political and Policy Sciences, the School of
Natural Sciences and Mathematics and the Erik Jonsson School of
Engineering and Computer Science jointly offer master's, doctoral and
graduate certificate programs in Geospatial Information Sciences.
Geospatial technology is a key component of Dr. Thuraisingham's broad
emphasis in knowledge discovery and security informatics, which
includes four professors and nearly 50 graduate students.
"About 50 percent of our work focuses on catching the bad guys' and 50
percent on protecting data," she said. "We are developing particularly
strong expertise in the semantic Web, data mining, data security and
social network analysis."
The team consisting of Dr. Thuraisingham, Dr. Khan and their computer
science colleagues Murat Kantarcioglu and Kevin Hamlen have brought in
more than $9 million in contracts and grants in the past four years,
including a prestigious National Science Foundation CAREER Award, a
U.S. Air Force Office of Scientific Research (AFOSR) Young Investigator
Research Program grant and a Department of Defense Multidisciplinary
University Research Initiative award as well as additional grants from
the NSF, the NGA, NASA, AFOSR, IARPA and the Office of Naval Research.
"We plan to expand our collaborations with additional companies and
agencies, including the Department of Homeland Security, the Defense
Advanced Research Projects Agency, the Army Research Office, the
National Institutes of Health, the Department of Energy and the
National Institute of Standards and Technology," Dr. Thuraisingham
Ed. note: This article
originally appeared on the University of Texas - Dallas News Center's
website on April 10, 2009, and is reprinted here with permission.