talks

Download

talks@lists.cs.princeton.edu

August 2020

1 participants
2 discussions

Anders Miltner will present his FPO "Synthesizing Lenses" on Friday, August 28, 2020 at 1pm via Zoom.
by Nicki Mahler 21 Aug '20

21 Aug '20

Anders Miltner will present his FPO "Synthesizing Lenses" on Friday, August 28, 2020 at 1pm via Zoom. Link to Zoom: [ https://princeton.zoom.us/j/6178355436 | https://princeton.zoom.us/j/6178355436 ] The members of his committee are as follows: David Walker (adviser); Readers: Zachary Kincaid and Kathleen Fisher (Tufts); Examiners: Aarti Gupta, Andrew Appel, and David Walker. A copy of his thesis, is available upon request. Please email ngotsis(a)cs.princeton if you would like a copy of the thesis. Everyone is invited to attend his talk. Abstract follows below. Lenses are bidirectional functions that satisfy a set of “round-tripping laws.” Lenses arise in a number of domains, and can implement serializer/deserializer pairs, parser/pretty printer pairs, and incremental data structure transformers. Domain-specific languages like Boomerang, Augeas, GRoundTram, BiFluX, BiYacc, Brul, BiGUL, and HOBiT allow programmers to write both transformations with a single program, and guarantee the transformations satisfy the round-tripping laws. However, writing in domain-specific lens languages can be hard, requiring the user to reason about fiddly details of the transformations, often within the context of a complex type system. In this work we introduce Optician, which provides an alternative method of developing lenses – synthesis. We develop a synthesis engine for Boomerang, a lens language for string transformations. Pairs of regular expressions serve as types for Boomerang lenses, which transform strings between the languages of those regular expressions. Instead of manually writing Boomerang lenses, programmers merely need to provide the type of the desired lens and a set of input/output examples describing the lens’s behavior. Optician will then synthesize a lens between the provided types that exhibits the demonstrated behavior. We demonstrate how to synthesize three classes of lenses: bijective lenses, quotient lenses, and symmetric lenses. Bijective lenses encode bijections, quotient lenses encode bijections modulo an equivalence relation, and symmetric lenses encode bidirectional transformations that may discard information when going from one format to another. We define a synthesizer that involves two cooperating procedures. One procedure proposes a candidate space of lenses, the second procedure searches through that space. To synthesize quotient lenses, regular expressions are augmented with equivalence information, and the synthesizer generates lenses that translate between the representative elements of the equivalence classes. To synthesize symmetric lenses, regular expressions are augmented with probability distributions, and the synthesis algorithm aims to avoid information loss. We evaluate Optician from a variety of benchmarks taken from the lens and synthesis literature, and find that we are able to synthesize all the lenses in our benchmark suite. We have integrated Optician into the Boomerang codebase, enabling users to either synthesize their lenses or write them by hand.

1 0

Arunesh Mathur will present his FPO "Identifying and measuring manipulative user interfaces at scale on the web" on Wednesday, August 26, 2020 at 2pm via Zoom.
by Nicki Mahler 21 Aug '20

21 Aug '20

Arunesh Mathur will present his FPO "Identifying and measuring manipulative user interfaces at scale on the web" on Wednesday, August 26, 2020 at 2pm via Zoom. Link to Zoom: [ https://princeton.zoom.us/j/92967251962 | https://princeton.zoom.us/j/92967251962 ] The members of his committee are as follows: Marshini Chetty (adviser); Readers: Marshini Chetty and Arvind Narayanan; Examiners: Jonathan Mayer, Brandon Stewart, and Janet Vertesi. A copy of his thesis, is available upon request. Please email ngotsis(a)cs.princeton if you would like a copy of the thesis. Everyone is invited to attend his talk. Abstract follows below. Powerful and otherwise trustworthy actors on the web gain from manipulating users and pushing them into making sub-optimal decisions. While prior work has documented examples of such manipulative practices, we lack a systematic understanding of their characteristics and their prevalence on the web. Building up this knowledge can lead to solutions that protect individuals and society from their harms. In this dissertation, I focus on manipulative practices that manifest in the user interface. I first describe the attributes of manipulative user interfaces. I show that these interfaces engineer users' choice architectures by either modifying the information available to users, or by modifying the set of choices available to users—eliminating and suppressing choices that disadvantage the manipulator. I then present the core contribution of this dissertation: automated methods that combine web automation and machine learning to identify manipulative interfaces at scale on the web. Using these methods, I conduct three measurements. First, I examine the extent to which content creators fail to disclose their endorsements on social media—misleading users into believing they are viewing unbiased, non-advertising content. Collecting and analyzing a dataset of 500K YouTube videos and 2 million Pinterest pins, I discover that ~90% of these endorsements go undisclosed. Second, I quantify the prevalence of dark patterns on shopping websites. Analyzing data I collected from 11K shopping websites, I discover 1,818 dark patterns on 1,254 websites that mislead, deceive, or coerce users into making more purchases or disclosing more information than they would otherwise. Finally, I quantify the prevalence of dark patterns and clickbait in political campaign emails. Collecting and analyzing a dataset of over 100K campaign emails in the U.S., I discover a long tail of candidates and organizations that use these interfaces to influence public opinion and to solicit donations. The median political campaign sends emails containing such interfaces with a probability of ~30%. I conclude with how the lessons learned from these measurements can be used to build technical defenses and to lay out policy recommendations to mitigate the spread of these interfaces

1 0