Amy X. Zhang
Massachusetts Institute of Technology
Network
Latest external collaboration on country level. Dive into details by clicking on the dots.
Publication
Featured researches published by Amy X. Zhang.
international conference on social computing | 2013
Amy X. Zhang; Anastasios Noulas; Salvatore Scellato; Cecilia Mascolo
Information garnered from activity on location-based social networks can be harnessed to characterize urban spaces and organize them into neighborhoods. We represent geographic points in the city using spatio-temporal information about Foursquare user check-ins and semantic information about places, with the goal of developing features to input into a novel neighborhood detection algorithm. The algorithm first employs a similarity metric that assesses the homogeneity of a geographic area, and then with a simple mechanism of geographic navigation, it detects the boundaries of a citys neighborhoods. The models and algorithms devised are subsequently integrated into a publicly available, map-based tool named Hood square that allows users to explore activities and neighborhoods in cities around the world. Finally, we evaluate Hood square in the context of are commendation application where user profiles are matched to urban neighborhoods. By comparing with a number of baselines, we demonstrate how Hood square can be used to accurately predict the home neighborhood of Twitter users. We also show that we are able to suggest neighborhoods geographically constrained in size, a desirable property in mobile recommendation scenarios for which geographical precision is key.
IEEE Journal of Selected Topics in Signal Processing | 2015
Salman Salamatian; Amy X. Zhang; Flávio du Pin Calmon; Sandilya Bhamidipati; Nadia Fawaz; Branislav Kveton; Pedro Oliveira; Nina Taft
We propose a practical methodology to protect a users private data, when he wishes to publicly release data that is correlated with his private data, to get some utility. Our approach relies on a general statistical inference framework that captures the privacy threat under inference attacks, given utility constraints. Under this framework, data is distorted before it is released, according to a probabilistic privacy mapping. This mapping is obtained by solving a convex optimization problem, which minimizes information leakage under a distortion constraint. We address practical challenges encountered when applying this theoretical framework to real world data. On one hand, the design of optimal privacy mappings requires knowledge of the prior distribution linking private data and data to be released, which is often unavailable in practice. On the other hand, the optimization may become untractable when data assumes values in large size alphabets, or is high dimensional. Our work makes three major contributions. First, we provide bounds on the impact of a mismatched prior on the privacy-utility tradeoff. Second, we show how to reduce the optimization size by introducing a quantization step, and how to generate privacy mappings under quantization. Third, we evaluate our method on two datasets, including a new dataset that we collected, showing correlations between political convictions and TV viewing habits. We demonstrate that good privacy properties can be achieved with limited distortion so as not to undermine the original purpose of the publicly released data, e.g., recommendations.
human factors in computing systems | 2015
Amy X. Zhang; Scott Counts
Social media has emerged as a prominent platform where people can express their feelings about social and political issues of our time. We study the many voices discussing an issue within a constituency and how they reflect ideology and may signal the outcome of important policy decisions. Focusing on the issue of same-sex marriage legalization, we examine almost 2 million public Twitter posts related to same-sex marriage in the U.S. states over the course of 4 years starting from 2011. Among other findings, we find evidence of moral culture wars between ideologies and show that constituencies that express higher levels of emotion and have fewer actively engaged participants often precede legalization efforts that fail. From our measures, we build statistical models to predict the outcome of potential policy changes, with our best model achieving 87% accuracy. We also achieve accuracies of 70%, comparable to public opinion surveys, many months before a policy decision. We discuss how these analyses can augment traditional political science techniques as well as assist activists and policy analysts in understanding discussions on important issues at a population scale.
user interface software and technology | 2014
Edward Benson; Amy X. Zhang; David R. Karger
Creating and publishing read-write-compute web applications requires programming skills beyond what most end users possess. But many end users know how to make spreadsheets that act as simple information management applications, some even with computation. We present a system for creating basic web applications using such spreadsheets in place of a server and using HTML to describe the client UI. Authors connect the two by placing spreadsheet references inside HTML attributes. Data computation is provided by spreadsheet formulas. The result is a reactive read-write-compute web page without a single line of Javascript code. Nearly all of the fifteen HTML novices we studied were able to connect HTML to spreadsheets using our method with minimal instruction. We draw conclusions from their experience and discuss future extensions to this programming model.
conference on computer supported cooperative work | 2017
Amy X. Zhang; Lea Verou; David R. Karger
Large-scale discussions between many participants abound on the internet today, on topics ranging from political arguments to group coordination. But as these discussions grow to tens of thousands of posts, they become ever more difficult for a reader to digest. In this article, we describe a workflow called recursive summarization, implemented in our Wikum prototype, that enables a large population of readers or editors to work in small doses to refine out the main points of the discussion. More than just a single summary, our workflow produces a summary tree that enables a reader to explore distinct subtopics at multiple levels of detail based on their interests. We describe lab evaluations showing that (i) Wikum can be used more effectively than a control to quickly construct a summary tree and (ii) the summary tree is more effective than the original discussion in helping readers identify and explore the main topics.
human factors in computing systems | 2016
Amy X. Zhang; Scott Counts
In the past few years an unprecedented wave of anti-abortion policies were introduced and enacted in state governments in the U.S., affecting millions of constituents. We study this rapid spread of policy change as a function of the underlying ideology of constituents. We examine over 200,000 public messages posted on Twitter surrounding abortion in the year 2013, a year that saw 82 new anti-abortion policies enacted. From these posts, we characterize peoples expressions of opinion on abortion and show how these expressions align with policy change on these issues. We detail a number of ideological differences between constituents in states enacting anti versus pro-abortion policies, such as a tension between the moral values of purity versus fairness, and a differing emphasis on the fetus versus the pregnant woman. We also find significant differences in how males versus females discuss the issue of abortion, including greater emphasis on health and religion by males. Using these measures to characterize states, we can construct models to explain the spread of abortion policy from state to state and project which types of abortion policies a state will introduce. Models defining state similarity using our Twitter-based measures improved policy projection accuracy by 7.32% and 12.02% on average over geographic and poll-based ideological similarity, respectively. Additionally, models constructed from the expressions of male-only constituents perform better than models from the expressions of female-only constituents, suggesting that the ideology of men is more aligned with the recent spread of anti-abortion legislation than that of women.
conference on computer supported cooperative work | 2016
Amy X. Zhang; Joshua M. Blum; David R. Karger
While the web contains many social websites, people are generally left in the dark about the activities of other people traversing the web as a whole. In this paper, we explore the potential benefits and privacy considerations around generating a real-time, publicly accessible stream of web activity where users can publish chosen parts of their web browsing data. Taking inspiration from social media systems, we describe individual benefits that can be unlocked by such sharing and that may incentivize users to publish aspects of their browsing. We ask whether and how these benefits outweigh potential costs in lost privacy. We conduct our study of public web activity sharing through scenario-based interviews and a field deployment of a tool for web activity sharing.
user interface software and technology | 2016
Lea Verou; Amy X. Zhang; David R. Karger
Many people can author static web pages with HTML and CSS but find it hard or impossible to program persistent, interactive web applications. We show that for a broad class of CRUD (Create, Read, Update, Delete) applications, this gap can be bridged. Mavo extends the declarative syntax of HTML to describe Web applications that manage, store and transform data. Using Mavo, authors with basic HTML knowledge define complex data schemas implicitly as they design their HTML layout. They need only add a few attributes and expressions to their HTML elements to transform their static design into a persistent, data-driven web application whose data can be edited by direct manipulation of the content in the browser. We evaluated Mavo with 20 users who marked up static designs---some provided by us, some their own creation---to transform them into fully functional web applications. Even users with no programming experience were able to quickly craft Mavo applications.
user interface software and technology | 2014
Juho Kim; Amy X. Zhang; Jihee Kim; Robert C. Miller; Krzysztof Z. Gajos
Long documents are abundant on the web today, and are accessed in increasing numbers from touchscreen devices such as mobile phones and tablets. Navigating long documents with small screens can be challenging both physically and cognitively because they compel the user to scroll a great deal and to mentally filter for important content. To support navigation of long documents on touchscreen devices, we introduce content-aware kinetic scrolling, a novel scrolling technique that dynamically applies pseudo-haptic feedback in the form of friction around points of high interest within the page. This allows users to quickly find interesting content while exploring without further cluttering the limited visual space. To model degrees of interest (DOI) for a variety of existing web pages, we introduce social wear, a method for capturing DOI based on social signals that indicate collective user interest. Our preliminary evaluation shows that users pay attention to items with kinetic scrolling feedback during search, recognition, and skimming tasks.
conference on information sciences and systems | 2013
Soheil Feizi; Amy X. Zhang; Muriel Médard
In this paper, by using network flow principles, we propose algorithms to address various challenges in cloud computing. One of the main challenges is to consider both communication and computation constraints in the network. In the proposed network flow framework, we model the amount of computation in each node of the network as a function of its total self-loop flows. We consider two computation cost models: a linear computation cost model and a maximum computation cost model. We show that, our network flow framework can be used as a systematic technique of balancing computation loads over different nodes of the network. This network flow framework can also be used for cloud network design. A network topology is optimal for certain computations if it maximizes the total computation rate under communication/computation constraints. We propose a greedy algorithm to design a cloud network with a certain network characteristics in terms of communication and computation costs. We provide simulation results to illustrate the performance of our algorithms.