Group items tagged

"A small research arm of the U.S. government's intelligence establishment wants to understand how speakers of Farsi, Russian, English, and Spanish see the world by building software that automatically evaluates their use of metaphors."

A new study finds that software designed to automatically read and grade essays can do as good a job as humans - maybe even better. The robo-readers are efficient, as well, and can grade 16,000 essays in 20 seconds. The caveats, however: computers don't do as well on comprehension, ascertaining facts vs. fiction, and assessing certain genres, like poetry.

"A small band of linguists, engineers and computer scientists, among others, are busy training computers to recognize hallmarks of what they call emotional speech - talk that reflects deception, anger, friendliness and even flirtation. ... Algorithms developed by Dr. Hirschberg and colleagues have been able to spot a liar 70 percent of the time in test situations, while people confronted with the same evidence had only 57 percent accuracy ... His lab has also found ways to use vocal cues to spot inebriation, though it hasn't yet had luck in making its computers detect humor - a hard task for the machines, he said."

This is a helpful site about how to read a spectrogram that hasn't been labelled, such as one that you might record yourself using special software.

gives overview of lots of websites and software available for language analysis

Youngsters and adults alike will benefit from a new tool designed to keep the Nlaka' pamux language alive and well in the Nicola Valley. The eight First Nations that make up the Citxw Nlaka' pamux Assembly (CNA) are each receiving six computers equipped with licenses to an online program designed to teach the language orally.

This article discusses the AI company Sanas' invention of technology that can mask "non-preferred" accents, intended for use in call centers. Controversy has erupted over the consequences of this technology, namely white-washing and the erasure of the culture and identity that is tied to accents.

Santiago Swallow is "a Mexican-born, American motivational speaker, consultant, educator, and author, whose speeches and publications focus on understanding modern culture in the age of social networking, globally interconnected media, user generated content and the Internet," and has "dedicated himself to helping others know more about how media and personality can manipulated in the 21st Century." Though completely fictional, he boasts a Wikipedia biography and a Twitter account with tens of thousands of followers. Making up-or at least "enhancing"-an identity like this is something real people do to increase their reputation, look popular, and sell themselves. There are equally real people who profit from this by selling fake followers created by software at the push of a button. Be afraid.

Interesting article on how reading books and embracing seemingly looked down upon majors like music, literature and such help when writing codes.

From a charred Hebrew scroll, researchers resurrected one of the earliest known versions of the Old Testament using a new digital reconstruction technique that may prove invaluable in revealing words from other previously unreadable finds, said scientists who plan to make the imaging software freely available.

This NYT article reports findings of a recently released study, which concluded that computers are capable of scoring essays on standardized tests as well, or possibly better, than humans. However, Les Perelman, a director of writing at MIT and a tester of the Educational Testing Services E-Rater program disagrees, arguing that the system can be easily "gamed".

Thanks to Jesse Huang for this article, which discusses the recent shooting and killing of Trayvon Martin by George Zimmerman, a Neighborhood Watch member, and the ensuing investigation as to whether the shooting was done in self-defense. Using software that examines characteristics like pitch and the space between spoken words to analyze voices, forensic audio experts are comparing a 911 call to a previous voice recording of Zimmerman and attempting to determine whether the background screams are that of Martin or Zimmerman. The article also includes discussion as to the reliability of this type of "voiceprint" analysis.

This is the newest version of WASP, now renamed. For Windows.

13 year old Trisha Prabhu of Naperville, IL, is a finalist in Google Science Fair 2014. Prabhu's project focuses on preventing cyber-bullying. Excerpted from her project summary statement: "Cyberbullying may result in depression, low self-esteem and in rare cases suicides in adolescent victims(12-18). Research shows that, over 50% of adolescents and teens have been bullied online and 10 to 20% experience it regularly. Research also shows that adolescents that post mean/hurtful messages may not understand the potential consequences of their actions because the pre-frontal cortex, the area of brain that controls reasoning and decision-making isn't developed until age 25. I hypothesized that if adolescents(ages 12-18) were provided an alert mechanism that suggested them to re-think their decision if they expressed willingness to post a mean/hurtful message on social media, the number of mean/hurtful messages adolescents will be willing to post would be lesser than adolescents that are not provided with such an alert mechanism. In order to check if my hypothesis was true, I created two Software systems: 1) Baseline 2) Rethink. "Rethink" system measured number of mean/hurtful messages adolescents were willing to post after being alerted to rethink, while the "Baseline" system measured the same without the alert. Results proved that adolescents were 93.43% less willing to post mean/hurtful messages using a "Rethink" system compared with "Baseline" system without alert."

Pew Researchers surveyed more than 1,500 technologists and scholars about the forces shaping the way people interact with one another online. They asked: "In the next decade, will public discourse online become more or less shaped by bad actors, harassment, trolls, and an overall tone of griping, distrust, and disgust?" The vast majority of techonolgists surveyed-81 percent of them-said they expect the tone of online discourse will either stay the same or get worse in the next decade. "Cyberattacks, doxing, and trolling will continue, while social platforms, security experts, ethicists, and others will wrangle over the best ways to balance security and privacy, freedom of speech, and user protections. A great deal of this will happen in public view," Susan Etlinger, a technology industry analyst, told Pew. "The more worrisome possibility is that privacy and safety advocates, in an effort to create a more safe and equal internet, will push bad actors into more-hidden channels such as Tor." Tor is software that enables people to browse and communicate online anonymously-so it's used by people who want to cover their tracks from government surveillance, those who want to access the dark web, trolls, whistleblowers, and others. The uncomfortable truth is that humans like trolling. It's easy for people to stay anonymous while they harass, pester, and bully other people online-and it's hard for platforms to design systems to stop them. Hard for two reasons: One, because of the "ever-expanding scale of internet discourse and its accelerating complexity," as Pew puts it. And, two, because technology companies seem to have little incentive to solve this problem for people. "Very often, hate, anxiety, and anger drive participation with the platform," said Frank Pasquale, a law professor at the University of Maryland, in the report. "Whatever behavior increases ad revenue will not only be permitted, but encouraged."