Association for Computing Machinery
Welcome to the December 23, 2011 edition of ACM TechNews, providing timely information for IT professionals three times a week.

Please Note: In observance of the Christmas holiday, TechNews will not be published on Monday, Dec. 26. Publication will resume Wednesday, Dec. 28.

ACM TechNews mobile apps are available for Android phones and tablets (click here) and for iPhones (click here) and iPads (click here).


Computer Scientists Create Algorithm That Measures Human Pecking Order
Technology Review (12/21/11)

Cornell University's Jon Kleinberg, who developed the Hyper Induced Topic Search (HITS) algorithm that led to Google's PageRank search algorithm, has developed a method for measuring power differences between individuals using the patterns of words they speak or write. "We show that in group discussions, power differentials between participants are subtly revealed by how much one individual immediately echoes the linguistic style of the person they are responding to," Kleinberg says. The key to the technique is linguistic co-ordination, in which speakers naturally copy the style of the interlocutors. The Cornell researchers focused on functional words that provide a grammatical framework for sentences but do not have real meaning themselves, such as articles, auxiliary verbs, conjunctions and high-frequency adverbs. The researchers studied editorial discussions between Wikipedia editors and transcripts of oral arguments in the U.S. Supreme Court. By looking at the changes in linguistic style that occur when people make the transition from non-admin to admin roles on Wikipedia, the researchers show that the pattern of linguistic co-ordination changes too. A similar effect occurs in the Supreme Court. "Our work is the first to identify connections between language coordination and social power relations at large scales, and across a diverse set of individuals and domains," Kleinberg says.

Traditional Social Networks Fueled Twitter's Spread
MIT News (12/21/11) Caroline McCall

Massachusetts Institute of Technology (MIT) researchers studied the growth of Twitter from 2006 to 2009 and found that its initial growth in the United States relied heavily on media attention and traditional social networks based on geographic proximity and socioeconomic similarity. The researchers examined data from 16,000 U.S. cities, focusing on the 408 with the highest number of Twitter users. They characterized cities as early adopters, early majority adopters, late majority adopters, or laggards based on when 13.5 percent of the population had Twitter accounts. "The social network needs geographical proximity," says MIT professor Marta Gonzalez. "In the U.S., anyway, space and similarity matter." The researchers found that Twitter's initially popularity was fueled by young, tech-savvy users in San Francisco and Boston. However, after that it began to follow a more traditional pattern by traveling short distances, which implied that personal interactions were important to its wider use. "Nobody has ever really looked at the diffusion among innovators of a no-risk, free, or low-cost product that’s only useful if other people join you," says MIT graduate student Jameson Toole. "It’s a new paradigm in economics: what to do with all these new things that are free and easy to share."

Fast Research Networks Make Global Collaboration Possible
Converge (12/21/11) Tanya Roscorla

While scientific research previously occurred at individual universities, with scientists traveling all over the world to share and compare data, new network services, such as Internet2, have made scientific collaboration much cheaper and easier. Internet2 created a private network that recently started upgrading to 100 gigabit capacity through Ciena. California Institute of Technology and University of Victoria researchers recently showcased a high-speed link between Seattle and Victoria, Australia that can transfer data at a combined rate of 186 gigabits per second. Now researchers are studying how to make file transfers more efficient through different types of protocols. Those research projects include the Large Hadron Collider, which is connected with the California Institute of Technology through a network of computers and research centers that work together on the particle physics data. Additionally, the Atacama Large Millimeter/submillimeter telescope array in Chile is expected to produce more than 100 gigabits per second of data in a few years, according to Internet2's executive director of network services Rob Vietzke. Internet2 allows university researchers to link remotely to the telescope in Chile. When networking in countries around the world, they also need local knowledge and expertise, so collaborating with people from those countries helps, Vietzke says. "It really is important that the campuses work together and they find that local knowledge because as we work around the world, it's really amazing what we take for granted here in terms of the capabilities we have," Vietzke says.

Trimming Time in the Stacks
UC Berkeley News Center (12/20/11) Nicole Freeling

University of California, Berkeley graduate student Aditi Muralidharan has developed WordSeer, a text-analyzing tool that she says could transform the process of scholarly research, reducing work that normally takes months to about five minutes. WordSeer uses a rubric about the way humans use and structure language to bring a facsimile of logic to the act of interpreting search results. WordSeer, which is based on natural language processing, features a user interface that enables people to pose questions of particular sets of text. "You don’t have to read all 4,000 matches to get a sense of the overall tone of the collection," Muralidharan says. Berkeley professors Bryan Wagner and Todd Carmody used WordSeer to research American slave narratives to better understand their relationship to God. "The WordSeer project is one of the most sophisticated tools for computational text analysis I've seen,” Wagner says. "There are many tools that tabulate words and phrases, but WordSeer can discern grammatical structure as well as stylistic features, representing the results in truly interactive visualizations."

Better Turbine Simulation Software to Yield Better Engines
Ohio Supercomputer Center (12/20/11) Kathryn Kelley; Jamie Abel

Ohio State University researchers are working to improve computational fluid dynamics (CFD) software that engineers use to simulate and evaluate the operation of turbomachinery. The researchers, led by Ohio State professor Jen-Ping Chen, are using the Ohio Supercomputing Center (OSC) to refine the CFD software as it confirms the flow field of engine components, focusing on high-pressure compressors and low-pressure turbines. "The world is demanding increasingly cleaner, more efficient and reliable power systems [and] it is essential that experts like Dr. Chen find innovative ways to improve the tools the engineers need to accomplish that goal, and we at OSC are proud to be able to provide the computational resources that make that effort successful," says OSC's Ashok Krishnamurthy. The simulation tools will provide engine designers with more physical insight to the complex flow field, which will lead to reduced testing and risk, faster time-to-market, and lower costs. "The successful combination of CFD simulation and experimentation can greatly supplement the understanding of fundamental fluid behavior of gas turbine systems, thus enhancing the ability of engineers to develop more advanced engine components," Chen says.

Chess Robots Have Trouble Grasping the Game
New Scientist (12/20/11) Duncan Graham-Rowe

Robots played chess during a competition in August, but the contest showed that humans still win hands down. During the event at the Association for the Advancement of Artificial Intelligence annual conference, robots were clumsy in manipulating physical objects. They had trouble moving pieces quickly, accurately, and in accordance with the rules of chess. Some robots used vision systems to determine the location of pieces, but rather than distinguish between a rook and a knight, for example, they remembered where pieces were last placed to identify them and then made moves. The vision systems often misread moves, which confused the bots as to what piece had been moved and where the others were on the board. A team led by the University of Albany's Mike Ferguson won the competition. The mobile robot, with an attached arm, used a vision system that was able to keep track of the board and pieces as moves were made, says the Georgia Institute of Technology's Mike Stilman. Still, the team's bot used a standard open source program for its chess tactics, and it would have provided little challenge for Deep Blue.

Patent Office Presents $50,000 Innovation Challenge
InformationWeek (12/20/11) Elizabeth Montalbano

The U.S. Trademark & Patent Office (USPTO) has attracted more than 900 applicants for its Innovation Challenge, a contest to develop new technology to accelerate the patent approval process. USPTO believes a new algorithm would help reduce the paper burden of annually processing 500,000 patent applications and has developed a challenge that asks image-processing specialists to work in teams of two over a month to develop an algorithm that can automatically identify and locate specific elements within patent documents. USPTO is particularly interested in finding a better way to examine illustrations that are key to whether a patent is approved or not. Information on illustrations is usually scattered on different pages of the document, and flipping back and forth between textual descriptions and drawings slows down patent examiners. "Few image-processing problems can be solved reliably today, but image-processing experts think this particular problem is sufficiently well defined to reap the benefits of crowdsourcing, which has a strong track record of surfacing novel solutions from unexpected places," says the U.S. Office of Science and Technology Policy's Robynn Sturm Steffen. Technologists are competing for a $50,000 prize, and the winner will be decided by Feb. 16, 2012.

Do You See What I See?
Los Alamos National Laboratory News (12/20/11) Nancy Ambrosiano

Researchers at Los Alamos National Laboratory (LANL), Chatham University, and Emory University have developed a model to teach a computer to identify an object simply by "looking" at an image, similar to the human brain, and to perform the visual task faster than humans. The computer model is based on human neural structure and function. The model leverages lateral connections between neurons in the same layer of a model of the human visual system, says Chatham University's Vadas Gintautas. The researchers demonstrated that the model qualitatively reproduces human performance on the same task, both in terms of time and difficulty. "Although this is certainly no guarantee that the human visual system is using lateral interactions in the same way to solve this task, it does open up a new way to approach object detection problems," Gintautas says. The researchers combined high-performance computer simulations of cortical circuits, using the neural simulation toolbox developed by LANL researchers called PetaVision, along with speed-of-sight psychophysical measurements of human contour perception.

Self-Healing Electronics Could Work Longer and Reduce Waste
University of Illinois News Bureau (12/20/11) Liz Ahlberg

University of Illinois researchers have developed a self-healing system that rejuvenates electrical conductivity to a damaged circuit in less than a second. "Rather than having to build in redundancies or to build in a sensory diagnostics system, this material is designed to take care of the problem itself," says Illinois professor Jeffrey Moore. The researchers placed tiny microcapsules on a gold line that acted as a circuit. If a crack develops, the microcapsules break open and release a liquid metal contained inside. The liquid metal fills in the gap in the circuit, restoring the electrical flow. "What’s really cool about this paper is it’s the first example of taking the microcapsule-based healing approach and applying it to a new function," says Illinois professor Scott White. The researchers have demonstrated that 90 percent of their samples retained 99 percent of original conductivity. The researchers plan to further refine the system and explore other possibilities for using microcapsules to control conductivity.

Virginia Tech's Wu Feng Unveils HokieSpeed, a New Powerful Supercomputer for the Masses
Virginia Tech News (12/19/11) Steven Mackay

Virginia Tech recently unveiled HokieSpeed, a new supercomputer that has a single-precision peak speed of 455 teraflops and a double-precision peak speed of 240 teraflops, fast enough to rank the system 96th on the most recent Top500 list. In addition, HokieSpeed's energy efficiency places it 11th on the most recent Green500 list, the highest ranked system in the United States. HokieSpeed contains 209 nodes, each of which consists of two six-core Intel central processing units and two NVIDIA 448-core graphical processing units. "This instrument will empower faculty, students, and staff across disciplines to tackle problems previously viewed as intractable or that required heroic efforts and significant domain-specific expertise to solve," says Virginia Tech professor Wu Feng. HokieSpeed was built for $1.4 million, the majority of which came from a $2 million U.S. National Science Foundation Major Research Instrumentation grant. HokieSpeed's visual wall will provide a 14-foot wide by four-foot tall display for end users to view their data. "What we want to do with HokieSpeed is to enable scientists to routinely do ‘what-if’ scenarios that they would not have been able to do or think of doing in the past," Feng says.

UVIC Team Infiltrates Paid 'Army' of Internet Posters
University of Victoria (12/19/2011) Patty Pitts

Consumers should think twice before basing their purchasing decisions on blogs and comments online, says University of Victoria computer science graduate student Cheng Chen. He studied the activities of online posters who are paid to hype products and services, and went undercover to investigate how Chinese technology companies Tencent and Qihoo 360 conduct marketing tactics on the Web sites and Cheng and Victoria professors Kui Wu and Venkatesh Srinivasan developed an approach that revealed the organizational structure of paid posters and found differences in the posting patterns of normal and paid posters. The researchers noticed a difference in the kinds of comments posted, percentage of replies, time of posts, the duration of activity, and the number of reports that received comments from posters. The popular Yahoo! Answers Web site could be the next target of Cheng's team. They also are considering developing an online service to help consumers identify potential paid posters. "Paying posters is popular, not only in China, but in every country," Cheng notes.
View Full Article - Link to Publication Homepage | Return to Headlines

Facebook Helps Researchers See How Friendships Form
National Science Foundation (12/19/11) Bobbie Mixon

Three Harvard University researchers used Facebook to study how people select friends and the role that friendship plays in transmitting tastes and new ideas. The researchers examined whether people become friends because they resemble one another or whether people become more like their friends over time. They found that people's individual tastes influence the formation of friendships much more than a person's individual pre-existing tastes spread through his or her friendships. "One feature of Internet relationships that is particularly amenable to our research question is the extent to which it fosters users to communicate their own taste preferences and consumption patterns," says Harvard's Jason Kaufman. The researchers found that students who like certain kinds of music and movies are more likely to become friends on Facebook, but the diffusion of tastes through friendship ties was very rare. However, the results challenge other research about the importance of peer influence. "The researchers may be finding so much peer influence because the kinds of models they are running aren't appropriate and they are misinterpreting their findings," says Harvard's Kevin Lewis. "Our research is unique in that it uses data on a complete social network of respondents--a cohort of students from the same college," Kaufman notes.

And the Winners of the Cyber Challenge Are...
Government Computer News (12/19/11) William Jackson

The SANS Institute and the Common Knowledge Scholarship Foundation recently announced the winners of the most recent U.S. Cyber Challenge, a competition to identify future cybersecurity professionals. Competitors came from more than 150 U.S. high schools, and the top finishers received more than $6,000 in scholarships. More significant than the performance of the winners is the fact that more than 2,000 students from 169 schools in 32 states and three territories participated, says (ICS)2 Foundation director Julie Peeler. “That’s a good number of kids who are being exposed to cybersecurity, regardless of whether they go into the profession,” Peeler says. “We are at least educating a larger workforce overall to an understanding of cybersecurity issues.” The Cyber Foundation competitions are part of a broader U.S. Cyber Challenge effort that includes summer camps for college, high school, and younger students, the Air Force Association's Cyber Patriot program, and several other scholarship and training programs aimed at developing a future professional cybersecurity workforce. Launched in 2009, the U.S. Cyber Challenge set a goal of identifying and recruiting 10,000 people with native skills needed for cybersecurity.

Abstract News © Copyright 2011 INFORMATION, INC.
Powered by Information, Inc.

To submit feedback about ACM TechNews, contact: [email protected]
Current ACM Members: Unsubscribe/Change your email subscription by logging in at myACM.
Non-Members: Unsubscribe