Fine-Grained Sentiment Analysis
Cynthia Thompson, University of San Francisco
03/05/2015
Targeted sentiment analysis expands on document or sentence-level polarity classification by identifying the sentiment expressed towards specific entities in a span of text. The task is challenging due to the large variety of such entities and to the fact that not all entities are relevant to sentiment analysis. I will describe our work applying supervised and semi-supervised learning techniques to performing fine-grained sentiment analysis.
Avoiding Pain With Persistent Data Structures
Bob Ippolito, Nom Labs
03/12/2015
The work we do with data structures has gotten a lot more sophisticated in recent years due to the rise of multicore devices, distributed applications, and the expectation that updates should be transactional and/or reversible. Many of the mutable data structures in common use do not adapt well to these constraints without excessive locking, copying, or other forms of bookkeeping. Persistent (effectively immutable) variants of data structures can often be used to simplify these classes of problems. While many of these data structures are already commonplace in functional programming languages such as Haskell, Scala or Erlang, they can be adapted for use in any garbage-collected language such as Java, Python or JavaScript. I will discuss the implementation and usage of several persistent data structures and compare them with their mutable counterparts.
Privacy Badger, Dnt, And A Web Without Tracking
Cooper Quintin, Electronic Frontier Foundation, San Francisco
03/26/2015
Cooper will introduce the design and implementation of Privacy Badger, EFF's new browser extension that automatically blocks both invisible trackers and spying ads. It is intended to be a minimal or zero-configuration option that most Internet users can use to prevent non-consensual third party collection of their reading habits from their everyday browser. Privacy Badger couples the recently developed HTTP Do Not Track opt out header with a number of heuristics for classifying the behavior of third parties, to automatically determine which should be blocked, which are needed but should have cookies blocked, and which are safe from a privacy perspective. Cooper will also talk about the current state of non-consensual tracking on the web, what methods are currently being used to track people and exploring what alternatives we can pursue.
Quantified Self, Augmented Reality And Psychological Modeling
Jason Shankel, The Stupid Fun Club (PIZZA AFTER TALK IN DARWIN 28)
04/02/2015
Let’s explore how the century-old technology of psychological modeling will impact the development of quantified self and augmented reality technologies in the century to come.
Stuxnet And The Age Of Digital Warfare
Kim Zetter, WIRED
04/09/2015
In June 2010, a small security firm in Belarus discovered a computer worm that had infected computers in Iran and was causing them to crash. The worm used an ingenious zero-day exploit to spread, but other than this it appeared to be generic malware designed for corporate espionage. But as digital detectives dug through the code and began to reverse-engineer its commands, they discovered it was much more sophisticated than previously believed and had a much more insidious goal— to physically sabotage equipment used in Iran's nuclear program. Stuxnet, as the malicious program was dubbed, was a landmark attack since it was the first cyberweapon ever discovered in the wild and was the first digital code to jump the gap from the digital world to the real world to cause physical destruction. This presentation focuses on how the brilliant attack was designed and unleashed on computers in Iran -- being targeted against five companies in Iran who could help the attackers reach their target -- how researchers discovered and deciphered it and how it's discovery led them to uncover an arsenal of espionage tools that were also created and unleashed by the same attackers. It will also examine how Stuxnet launched a new era of warfare and how critical infrastructure systems in the U.S. and elsewhere are now at risk of ‘blowback’ and copycat attacks thanks to the authors of Stuxnet.