Search | arXiv e-print repository

<!DOCTYPE html> <html lang="en"> <head> <meta charset="utf-8"/> <meta name="viewport" content="width=device-width, initial-scale=1"/>  <link rel="apple-touch-icon" sizes="180x180" href="https://static.arxiv.org/static/base/1.0.0a5/images/icons/apple-touch-icon.png"> <link rel="icon" type="image/png" sizes="32x32" href="https://static.arxiv.org/static/base/1.0.0a5/images/icons/favicon-32x32.png"> <link rel="icon" type="image/png" sizes="16x16" href="https://static.arxiv.org/static/base/1.0.0a5/images/icons/favicon-16x16.png"> <link rel="manifest" href="https://static.arxiv.org/static/base/1.0.0a5/images/icons/site.webmanifest"> <link rel="mask-icon" href="https://static.arxiv.org/static/base/1.0.0a5/images/icons/safari-pinned-tab.svg" color="#b31b1b"> <link rel="shortcut icon" href="https://static.arxiv.org/static/base/1.0.0a5/images/icons/favicon.ico"> <meta name="msapplication-TileColor" content="#b31b1b"> <meta name="msapplication-config" content="images/icons/browserconfig.xml"> <meta name="theme-color" content="#b31b1b">  <title>Search | arXiv e-print repository</title> <script defer src="https://static.arxiv.org/static/base/1.0.0a5/fontawesome-free-5.11.2-web/js/all.js"></script> <link rel="stylesheet" href="https://static.arxiv.org/static/base/1.0.0a5/css/arxivstyle.css" /> <script type="text/x-mathjax-config"> MathJax.Hub.Config({ messageStyle: "none", extensions: ["tex2jax.js"], jax: ["input/TeX", "output/HTML-CSS"], tex2jax: { inlineMath: [ ['$','$'], ["\$","\$"] ], displayMath: [ ['$$','$$'], ["\\[","\\]"] ], processEscapes: true, ignoreClass: '.*', processClass: 'mathjax.*' }, TeX: { extensions: ["AMSmath.js", "AMSsymbols.js", "noErrors.js"], noErrors: { inlineDelimiters: ["$","$"], multiLine: false, style: { "font-size": "normal", "border": "" } } }, "HTML-CSS": { availableFonts: ["TeX"] } }); </script> <script src='//static.arxiv.org/MathJax-2.7.3/MathJax.js'></script> <script src="https://static.arxiv.org/static/base/1.0.0a5/js/notification.js"></script> <link rel="stylesheet" href="https://static.arxiv.org/static/search/0.5.6/css/bulma-tooltip.min.css" /> <link rel="stylesheet" href="https://static.arxiv.org/static/search/0.5.6/css/search.css" /> <script src="https://code.jquery.com/jquery-3.2.1.slim.min.js" integrity="sha256-k2WSCIexGzOj3Euiig+TlR8gA0EmPjuc79OEeY5L45g=" crossorigin="anonymous"></script> <script src="https://static.arxiv.org/static/search/0.5.6/js/fieldset.js"></script> <style> radio#cf-customfield_11400 { display: none; } </style> </head> <body> <header><a href="#main-container" class="is-sr-only">Skip to main content</a>  <div class="attribution level is-marginless" role="banner"> <div class="level-left"> <a class="level-item" href="https://cornell.edu/"><img src="https://static.arxiv.org/static/base/1.0.0a5/images/cornell-reduced-white-SMALL.svg" alt="Cornell University" width="200" aria-label="logo" /></a> </div> <div class="level-right is-marginless"><p class="sponsors level-item is-marginless"><span id="support-ack-url">We gratefully acknowledge support from<br /> the Simons Foundation, <a href="https://info.arxiv.org/about/ourmembers.html">member institutions</a>, and all contributors. <a href="https://info.arxiv.org/about/donate.html">Donate</a></span></p></div> </div>  <div class="identity level is-marginless"> <div class="level-left"> <div class="level-item"> <a class="arxiv" href="https://arxiv.org/" aria-label="arxiv-logo"> <img src="https://static.arxiv.org/static/base/1.0.0a5/images/arxiv-logo-one-color-white.svg" aria-label="logo" alt="arxiv logo" width="85" style="width:85px;"/> </a> </div> </div> <div class="search-block level-right"> <form class="level-item mini-search" method="GET" action="https://arxiv.org/search"> <div class="field has-addons"> <div class="control"> <input class="input is-small" type="text" name="query" placeholder="Search..." aria-label="Search term or terms" /> <p class="help"><a href="https://info.arxiv.org/help">Help</a> | <a href="https://arxiv.org/search/advanced">Advanced Search</a></p> </div> <div class="control"> <div class="select is-small"> <select name="searchtype" aria-label="Field to search"> <option value="all" selected="selected">All fields</option> <option value="title">Title</option> <option value="author">Author</option> <option value="abstract">Abstract</option> <option value="comments">Comments</option> <option value="journal_ref">Journal reference</option> <option value="acm_class">ACM classification</option> <option value="msc_class">MSC classification</option> <option value="report_num">Report number</option> <option value="paper_id">arXiv identifier</option> <option value="doi">DOI</option> <option value="orcid">ORCID</option> <option value="author_id">arXiv author ID</option> <option value="help">Help pages</option> <option value="full_text">Full text</option> </select> </div> </div> <input type="hidden" name="source" value="header"> <button class="button is-small is-cul-darker">Search</button> </div> </form> </div> </div>  <div class="container"> <div class="user-tools is-size-7 has-text-right has-text-weight-bold" role="navigation" aria-label="User menu"> <a href="https://arxiv.org/login">Login</a> </div> </div> </header> <main class="container" id="main-container"> <div class="level is-marginless"> <div class="level-left"> <h1 class="title is-clearfix"> Showing 1–20 of 20 results for author: <span class="mathjax">Thain, N</span> </h1> </div> <div class="level-right is-hidden-mobile">  <span class="help" style="display: inline-block;"><a href="https://github.com/arXiv/arxiv-search/releases">Search v0.5.6 released 2020-02-24</a>  </span> </div> </div> <div class="content"> <form method="GET" action="/search/cs" aria-role="search"> Searching in archive <strong>cs</strong>. <a href="/search/?searchtype=author&query=Thain%2C+N">Search in all archives.</a> <div class="field has-addons-tablet"> <div class="control is-expanded"> <label for="query" class="hidden-label">Search term or terms</label> <input class="input is-medium" id="query" name="query" placeholder="Search term..." type="text" value="Thain, N"> </div> <div class="select control is-medium"> <label class="is-hidden" for="searchtype">Field</label> <select class="is-medium" id="searchtype" name="searchtype"><option value="all">All fields</option><option value="title">Title</option><option selected value="author">Author(s)</option><option value="abstract">Abstract</option><option value="comments">Comments</option><option value="journal_ref">Journal reference</option><option value="acm_class">ACM classification</option><option value="msc_class">MSC classification</option><option value="report_num">Report number</option><option value="paper_id">arXiv identifier</option><option value="doi">DOI</option><option value="orcid">ORCID</option><option value="license">License (URI)</option><option value="author_id">arXiv author ID</option><option value="help">Help pages</option><option value="full_text">Full text</option></select> </div> <div class="control"> <button class="button is-link is-medium">Search</button> </div> </div> <div class="field"> <div class="control is-size-7"> <label class="radio"> <input checked id="abstracts-0" name="abstracts" type="radio" value="show"> Show abstracts </label> <label class="radio"> <input id="abstracts-1" name="abstracts" type="radio" value="hide"> Hide abstracts </label> </div> </div> <div class="is-clearfix" style="height: 2.5em"> <div class="is-pulled-right"> <a href="/search/advanced?terms-0-term=Thain%2C+N&terms-0-field=author&size=50&order=-announced_date_first">Advanced Search</a> </div> </div> <input type="hidden" name="order" value="-announced_date_first"> <input type="hidden" name="size" value="50"> </form> <div class="level breathe-horizontal"> <div class="level-left"> <form method="GET" action="/search/"> <div style="display: none;"> <select id="searchtype" name="searchtype"><option value="all">All fields</option><option value="title">Title</option><option selected value="author">Author(s)</option><option value="abstract">Abstract</option><option value="comments">Comments</option><option value="journal_ref">Journal reference</option><option value="acm_class">ACM classification</option><option value="msc_class">MSC classification</option><option value="report_num">Report number</option><option value="paper_id">arXiv identifier</option><option value="doi">DOI</option><option value="orcid">ORCID</option><option value="license">License (URI)</option><option value="author_id">arXiv author ID</option><option value="help">Help pages</option><option value="full_text">Full text</option></select> <input id="query" name="query" type="text" value="Thain, N"> <ul id="abstracts"><li><input checked id="abstracts-0" name="abstracts" type="radio" value="show"> <label for="abstracts-0">Show abstracts</label></li><li><input id="abstracts-1" name="abstracts" type="radio" value="hide"> <label for="abstracts-1">Hide abstracts</label></li></ul> </div> <div class="box field is-grouped is-grouped-multiline level-item"> <div class="control"> <span class="select is-small"> <select id="size" name="size"><option value="25">25</option><option selected value="50">50</option><option value="100">100</option><option value="200">200</option></select> </span> <label for="size">results per page</label>. </div> <div class="control"> <label for="order">Sort results by</label> <span class="select is-small"> <select id="order" name="order"><option selected value="-announced_date_first">Announcement date (newest first)</option><option value="announced_date_first">Announcement date (oldest first)</option><option value="-submitted_date">Submission date (newest first)</option><option value="submitted_date">Submission date (oldest first)</option><option value="">Relevance</option></select> </span> </div> <div class="control"> <button class="button is-small is-link">Go</button> </div> </div> </form> </div> </div> <ol class="breathe-horizontal" start="1"> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/2403.08904">arXiv:2403.08904</a> <span> [<a href="https://arxiv.org/pdf/2403.08904">pdf</a>, <a href="https://arxiv.org/format/2403.08904">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Computation and Language">cs.CL</span> </div> </div> <p class="title is-5 mathjax"> Detecting Hallucination and Coverage Errors in Retrieval Augmented Generation for Controversial Topics </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Chang%2C+T+A">Tyler A. Chang</a>, <a href="/search/cs?searchtype=author&query=Tomanek%2C+K">Katrin Tomanek</a>, <a href="/search/cs?searchtype=author&query=Hoffmann%2C+J">Jessica Hoffmann</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=van+Liemt%2C+E">Erin van Liemt</a>, <a href="/search/cs?searchtype=author&query=Meier-Hellstern%2C+K">Kathleen Meier-Hellstern</a>, <a href="/search/cs?searchtype=author&query=Dixon%2C+L">Lucas Dixon</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="2403.08904v1-abstract-short" style="display: inline;"> We explore a strategy to handle controversial topics in LLM-based chatbots based on Wikipedia's Neutral Point of View (NPOV) principle: acknowledge the absence of a single true answer and surface multiple perspectives. We frame this as retrieval augmented generation, where perspectives are retrieved from a knowledge base and the LLM is tasked with generating a fluent and faithful response from the… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2403.08904v1-abstract-full').style.display = 'inline'; document.getElementById('2403.08904v1-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="2403.08904v1-abstract-full" style="display: none;"> We explore a strategy to handle controversial topics in LLM-based chatbots based on Wikipedia's Neutral Point of View (NPOV) principle: acknowledge the absence of a single true answer and surface multiple perspectives. We frame this as retrieval augmented generation, where perspectives are retrieved from a knowledge base and the LLM is tasked with generating a fluent and faithful response from the given perspectives. As a starting point, we use a deterministic retrieval system and then focus on common LLM failure modes that arise during this approach to text generation, namely hallucination and coverage errors. We propose and evaluate three methods to detect such errors based on (1) word-overlap, (2) salience, and (3) LLM-based classifiers. Our results demonstrate that LLM-based classifiers, even when trained only on synthetic errors, achieve high error detection performance, with ROC AUC scores of 95.3% for hallucination and 90.5% for coverage error detection on unambiguous error cases. We show that when no training data is available, our other methods still yield good results on hallucination (84.0%) and coverage error (85.2%) detection. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2403.08904v1-abstract-full').style.display = 'none'; document.getElementById('2403.08904v1-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 13 March, 2024; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> March 2024. </p> <p class="comments is-size-7"> <span class="has-text-black-bis has-text-weight-semibold">Comments:</span> <span class="has-text-grey-dark mathjax">Accepted at LREC-COLING 2024</span> </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/2403.08295">arXiv:2403.08295</a> <span> [<a href="https://arxiv.org/pdf/2403.08295">pdf</a>, <a href="https://arxiv.org/format/2403.08295">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Computation and Language">cs.CL</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Artificial Intelligence">cs.AI</span> </div> </div> <p class="title is-5 mathjax"> Gemma: Open Models Based on Gemini Research and Technology </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Gemma+Team"> Gemma Team</a>, <a href="/search/cs?searchtype=author&query=Mesnard%2C+T">Thomas Mesnard</a>, <a href="/search/cs?searchtype=author&query=Hardin%2C+C">Cassidy Hardin</a>, <a href="/search/cs?searchtype=author&query=Dadashi%2C+R">Robert Dadashi</a>, <a href="/search/cs?searchtype=author&query=Bhupatiraju%2C+S">Surya Bhupatiraju</a>, <a href="/search/cs?searchtype=author&query=Pathak%2C+S">Shreya Pathak</a>, <a href="/search/cs?searchtype=author&query=Sifre%2C+L">Laurent Sifre</a>, <a href="/search/cs?searchtype=author&query=Rivi%C3%A8re%2C+M">Morgane Rivi猫re</a>, <a href="/search/cs?searchtype=author&query=Kale%2C+M+S">Mihir Sanjay Kale</a>, <a href="/search/cs?searchtype=author&query=Love%2C+J">Juliette Love</a>, <a href="/search/cs?searchtype=author&query=Tafti%2C+P">Pouya Tafti</a>, <a href="/search/cs?searchtype=author&query=Hussenot%2C+L">L茅onard Hussenot</a>, <a href="/search/cs?searchtype=author&query=Sessa%2C+P+G">Pier Giuseppe Sessa</a>, <a href="/search/cs?searchtype=author&query=Chowdhery%2C+A">Aakanksha Chowdhery</a>, <a href="/search/cs?searchtype=author&query=Roberts%2C+A">Adam Roberts</a>, <a href="/search/cs?searchtype=author&query=Barua%2C+A">Aditya Barua</a>, <a href="/search/cs?searchtype=author&query=Botev%2C+A">Alex Botev</a>, <a href="/search/cs?searchtype=author&query=Castro-Ros%2C+A">Alex Castro-Ros</a>, <a href="/search/cs?searchtype=author&query=Slone%2C+A">Ambrose Slone</a>, <a href="/search/cs?searchtype=author&query=H%C3%A9liou%2C+A">Am茅lie H茅liou</a>, <a href="/search/cs?searchtype=author&query=Tacchetti%2C+A">Andrea Tacchetti</a>, <a href="/search/cs?searchtype=author&query=Bulanova%2C+A">Anna Bulanova</a>, <a href="/search/cs?searchtype=author&query=Paterson%2C+A">Antonia Paterson</a>, <a href="/search/cs?searchtype=author&query=Tsai%2C+B">Beth Tsai</a>, <a href="/search/cs?searchtype=author&query=Shahriari%2C+B">Bobak Shahriari</a> , et al. (83 additional authors not shown) </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="2403.08295v4-abstract-short" style="display: inline;"> This work introduces Gemma, a family of lightweight, state-of-the art open models built from the research and technology used to create Gemini models. Gemma models demonstrate strong performance across academic benchmarks for language understanding, reasoning, and safety. We release two sizes of models (2 billion and 7 billion parameters), and provide both pretrained and fine-tuned checkpoints. Ge… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2403.08295v4-abstract-full').style.display = 'inline'; document.getElementById('2403.08295v4-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="2403.08295v4-abstract-full" style="display: none;"> This work introduces Gemma, a family of lightweight, state-of-the art open models built from the research and technology used to create Gemini models. Gemma models demonstrate strong performance across academic benchmarks for language understanding, reasoning, and safety. We release two sizes of models (2 billion and 7 billion parameters), and provide both pretrained and fine-tuned checkpoints. Gemma outperforms similarly sized open models on 11 out of 18 text-based tasks, and we present comprehensive evaluations of safety and responsibility aspects of the models, alongside a detailed description of model development. We believe the responsible release of LLMs is critical for improving the safety of frontier models, and for enabling the next wave of LLM innovations. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2403.08295v4-abstract-full').style.display = 'none'; document.getElementById('2403.08295v4-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 16 April, 2024; <span class="has-text-black-bis has-text-weight-semibold">v1</span> submitted 13 March, 2024; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> March 2024. </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/2403.04894">arXiv:2403.04894</a> <span> [<a href="https://arxiv.org/pdf/2403.04894">pdf</a>, <a href="https://arxiv.org/format/2403.04894">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Computation and Language">cs.CL</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Artificial Intelligence">cs.AI</span> </div> </div> <p class="title is-5 mathjax"> ConstitutionalExperts: Training a Mixture of Principle-based Prompts </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Petridis%2C+S">Savvas Petridis</a>, <a href="/search/cs?searchtype=author&query=Wedin%2C+B">Ben Wedin</a>, <a href="/search/cs?searchtype=author&query=Yuan%2C+A">Ann Yuan</a>, <a href="/search/cs?searchtype=author&query=Wexler%2C+J">James Wexler</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="2403.04894v1-abstract-short" style="display: inline;"> Large language models (LLMs) are highly capable at a variety of tasks given the right prompt, but writing one is still a difficult and tedious process. In this work, we introduce ConstitutionalExperts, a method for learning a prompt consisting of constitutional principles (i.e. rules), given a training dataset. Unlike prior methods that optimize the prompt as a single entity, our method incrementa… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2403.04894v1-abstract-full').style.display = 'inline'; document.getElementById('2403.04894v1-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="2403.04894v1-abstract-full" style="display: none;"> Large language models (LLMs) are highly capable at a variety of tasks given the right prompt, but writing one is still a difficult and tedious process. In this work, we introduce ConstitutionalExperts, a method for learning a prompt consisting of constitutional principles (i.e. rules), given a training dataset. Unlike prior methods that optimize the prompt as a single entity, our method incrementally improves the prompt by surgically editing individual principles. We also show that we can improve overall performance by learning unique prompts for different semantic regions of the training data and using a mixture-of-experts (MoE) architecture to route inputs at inference time. We compare our method to other state of the art prompt-optimization techniques across six benchmark datasets. We also investigate whether MoE improves these other techniques. Our results suggest that ConstitutionalExperts outperforms other prompt optimization techniques by 10.9% (F1) and that mixture-of-experts improves all techniques, suggesting its broad applicability. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2403.04894v1-abstract-full').style.display = 'none'; document.getElementById('2403.04894v1-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 7 March, 2024; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> March 2024. </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/2305.13535">arXiv:2305.13535</a> <span> [<a href="https://arxiv.org/pdf/2305.13535">pdf</a>, <a href="https://arxiv.org/format/2305.13535">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Computation and Language">cs.CL</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Machine Learning">cs.LG</span> </div> </div> <p class="title is-5 mathjax"> Improving Classifier Robustness through Active Generation of Pairwise Counterfactuals </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Balashankar%2C+A">Ananth Balashankar</a>, <a href="/search/cs?searchtype=author&query=Wang%2C+X">Xuezhi Wang</a>, <a href="/search/cs?searchtype=author&query=Qin%2C+Y">Yao Qin</a>, <a href="/search/cs?searchtype=author&query=Packer%2C+B">Ben Packer</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Chen%2C+J">Jilin Chen</a>, <a href="/search/cs?searchtype=author&query=Chi%2C+E+H">Ed H. Chi</a>, <a href="/search/cs?searchtype=author&query=Beutel%2C+A">Alex Beutel</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="2305.13535v1-abstract-short" style="display: inline;"> Counterfactual Data Augmentation (CDA) is a commonly used technique for improving robustness in natural language classifiers. However, one fundamental challenge is how to discover meaningful counterfactuals and efficiently label them, with minimal human labeling cost. Most existing methods either completely rely on human-annotated labels, an expensive process which limits the scale of counterfactu… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2305.13535v1-abstract-full').style.display = 'inline'; document.getElementById('2305.13535v1-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="2305.13535v1-abstract-full" style="display: none;"> Counterfactual Data Augmentation (CDA) is a commonly used technique for improving robustness in natural language classifiers. However, one fundamental challenge is how to discover meaningful counterfactuals and efficiently label them, with minimal human labeling cost. Most existing methods either completely rely on human-annotated labels, an expensive process which limits the scale of counterfactual data, or implicitly assume label invariance, which may mislead the model with incorrect labels. In this paper, we present a novel framework that utilizes counterfactual generative models to generate a large number of diverse counterfactuals by actively sampling from regions of uncertainty, and then automatically label them with a learned pairwise classifier. Our key insight is that we can more correctly label the generated counterfactuals by training a pairwise classifier that interpolates the relationship between the original example and the counterfactual. We demonstrate that with a small amount of human-annotated counterfactual data (10%), we can generate a counterfactual augmentation dataset with learned labels, that provides an 18-20% improvement in robustness and a 14-21% reduction in errors on 6 out-of-domain datasets, comparable to that of a fully human-annotated counterfactual dataset for both sentiment classification and question paraphrase tasks. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2305.13535v1-abstract-full').style.display = 'none'; document.getElementById('2305.13535v1-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 22 May, 2023; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> May 2023. </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/2302.06598">arXiv:2302.06598</a> <span> [<a href="https://arxiv.org/pdf/2302.06598">pdf</a>, <a href="https://arxiv.org/format/2302.06598">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Computation and Language">cs.CL</span> </div> </div> <p class="title is-5 mathjax"> Gradient-Based Automated Iterative Recovery for Parameter-Efficient Tuning </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Mozes%2C+M">Maximilian Mozes</a>, <a href="/search/cs?searchtype=author&query=Bolukbasi%2C+T">Tolga Bolukbasi</a>, <a href="/search/cs?searchtype=author&query=Yuan%2C+A">Ann Yuan</a>, <a href="/search/cs?searchtype=author&query=Liu%2C+F">Frederick Liu</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Dixon%2C+L">Lucas Dixon</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="2302.06598v1-abstract-short" style="display: inline;"> Pretrained large language models (LLMs) are able to solve a wide variety of tasks through transfer learning. Various explainability methods have been developed to investigate their decision making process. TracIn (Pruthi et al., 2020) is one such gradient-based method which explains model inferences based on the influence of training examples. In this paper, we explore the use of TracIn to improve… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2302.06598v1-abstract-full').style.display = 'inline'; document.getElementById('2302.06598v1-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="2302.06598v1-abstract-full" style="display: none;"> Pretrained large language models (LLMs) are able to solve a wide variety of tasks through transfer learning. Various explainability methods have been developed to investigate their decision making process. TracIn (Pruthi et al., 2020) is one such gradient-based method which explains model inferences based on the influence of training examples. In this paper, we explore the use of TracIn to improve model performance in the parameter-efficient tuning (PET) setting. We develop conversational safety classifiers via the prompt-tuning PET method and show how the unique characteristics of the PET regime enable TracIn to identify the cause for certain misclassifications by LLMs. We develop a new methodology for using gradient-based explainability techniques to improve model performance, G-BAIR: gradient-based automated iterative recovery. We show that G-BAIR can recover LLM performance on benchmarks after manually corrupting training labels. This suggests that influence methods like TracIn can be used to automatically perform data cleaning, and introduces the potential for interactive debugging and relabeling for PET-based transfer learning methods. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2302.06598v1-abstract-full').style.display = 'none'; document.getElementById('2302.06598v1-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 13 February, 2023; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> February 2023. </p> <p class="comments is-size-7"> <span class="has-text-black-bis has-text-weight-semibold">Comments:</span> <span class="has-text-grey-dark mathjax">Pre-print</span> </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/2302.06541">arXiv:2302.06541</a> <span> [<a href="https://arxiv.org/pdf/2302.06541">pdf</a>, <a href="https://arxiv.org/format/2302.06541">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Computation and Language">cs.CL</span> </div> </div> <p class="title is-5 mathjax"> Towards Agile Text Classifiers for Everyone </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Mozes%2C+M">Maximilian Mozes</a>, <a href="/search/cs?searchtype=author&query=Hoffmann%2C+J">Jessica Hoffmann</a>, <a href="/search/cs?searchtype=author&query=Tomanek%2C+K">Katrin Tomanek</a>, <a href="/search/cs?searchtype=author&query=Kouate%2C+M">Muhamed Kouate</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Yuan%2C+A">Ann Yuan</a>, <a href="/search/cs?searchtype=author&query=Bolukbasi%2C+T">Tolga Bolukbasi</a>, <a href="/search/cs?searchtype=author&query=Dixon%2C+L">Lucas Dixon</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="2302.06541v2-abstract-short" style="display: inline;"> Text-based safety classifiers are widely used for content moderation and increasingly to tune generative language model behavior - a topic of growing concern for the safety of digital assistants and chatbots. However, different policies require different classifiers, and safety policies themselves improve from iteration and adaptation. This paper introduces and evaluates methods for agile text cla… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2302.06541v2-abstract-full').style.display = 'inline'; document.getElementById('2302.06541v2-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="2302.06541v2-abstract-full" style="display: none;"> Text-based safety classifiers are widely used for content moderation and increasingly to tune generative language model behavior - a topic of growing concern for the safety of digital assistants and chatbots. However, different policies require different classifiers, and safety policies themselves improve from iteration and adaptation. This paper introduces and evaluates methods for agile text classification, whereby classifiers are trained using small, targeted datasets that can be quickly developed for a particular policy. Experimenting with 7 datasets from three safety-related domains, comprising 15 annotation schemes, led to our key finding: prompt-tuning large language models, like PaLM 62B, with a labeled dataset of as few as 80 examples can achieve state-of-the-art performance. We argue that this enables a paradigm shift for text classification, especially for models supporting safer online discourse. Instead of collecting millions of examples to attempt to create universal safety classifiers over months or years, classifiers could be tuned using small datasets, created by individuals or small organizations, tailored for specific use cases, and iterated on and adapted in the time-span of a day. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2302.06541v2-abstract-full').style.display = 'none'; document.getElementById('2302.06541v2-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 21 October, 2023; <span class="has-text-black-bis has-text-weight-semibold">v1</span> submitted 13 February, 2023; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> February 2023. </p> <p class="comments is-size-7"> <span class="has-text-black-bis has-text-weight-semibold">Comments:</span> <span class="has-text-grey-dark mathjax">Findings of EMNLP 2023</span> </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/2207.07411">arXiv:2207.07411</a> <span> [<a href="https://arxiv.org/pdf/2207.07411">pdf</a>, <a href="https://arxiv.org/format/2207.07411">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Machine Learning">cs.LG</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Machine Learning">stat.ML</span> </div> </div> <p class="title is-5 mathjax"> Plex: Towards Reliability using Pretrained Large Model Extensions </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Tran%2C+D">Dustin Tran</a>, <a href="/search/cs?searchtype=author&query=Liu%2C+J">Jeremiah Liu</a>, <a href="/search/cs?searchtype=author&query=Dusenberry%2C+M+W">Michael W. Dusenberry</a>, <a href="/search/cs?searchtype=author&query=Phan%2C+D">Du Phan</a>, <a href="/search/cs?searchtype=author&query=Collier%2C+M">Mark Collier</a>, <a href="/search/cs?searchtype=author&query=Ren%2C+J">Jie Ren</a>, <a href="/search/cs?searchtype=author&query=Han%2C+K">Kehang Han</a>, <a href="/search/cs?searchtype=author&query=Wang%2C+Z">Zi Wang</a>, <a href="/search/cs?searchtype=author&query=Mariet%2C+Z">Zelda Mariet</a>, <a href="/search/cs?searchtype=author&query=Hu%2C+H">Huiyi Hu</a>, <a href="/search/cs?searchtype=author&query=Band%2C+N">Neil Band</a>, <a href="/search/cs?searchtype=author&query=Rudner%2C+T+G+J">Tim G. J. Rudner</a>, <a href="/search/cs?searchtype=author&query=Singhal%2C+K">Karan Singhal</a>, <a href="/search/cs?searchtype=author&query=Nado%2C+Z">Zachary Nado</a>, <a href="/search/cs?searchtype=author&query=van+Amersfoort%2C+J">Joost van Amersfoort</a>, <a href="/search/cs?searchtype=author&query=Kirsch%2C+A">Andreas Kirsch</a>, <a href="/search/cs?searchtype=author&query=Jenatton%2C+R">Rodolphe Jenatton</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Yuan%2C+H">Honglin Yuan</a>, <a href="/search/cs?searchtype=author&query=Buchanan%2C+K">Kelly Buchanan</a>, <a href="/search/cs?searchtype=author&query=Murphy%2C+K">Kevin Murphy</a>, <a href="/search/cs?searchtype=author&query=Sculley%2C+D">D. Sculley</a>, <a href="/search/cs?searchtype=author&query=Gal%2C+Y">Yarin Gal</a>, <a href="/search/cs?searchtype=author&query=Ghahramani%2C+Z">Zoubin Ghahramani</a>, <a href="/search/cs?searchtype=author&query=Snoek%2C+J">Jasper Snoek</a> , et al. (1 additional authors not shown) </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="2207.07411v1-abstract-short" style="display: inline;"> A recent trend in artificial intelligence is the use of pretrained models for language and vision tasks, which have achieved extraordinary performance but also puzzling failures. Probing these models' abilities in diverse ways is therefore critical to the field. In this paper, we explore the reliability of models, where we define a reliable model as one that not only achieves strong predictive per… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2207.07411v1-abstract-full').style.display = 'inline'; document.getElementById('2207.07411v1-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="2207.07411v1-abstract-full" style="display: none;"> A recent trend in artificial intelligence is the use of pretrained models for language and vision tasks, which have achieved extraordinary performance but also puzzling failures. Probing these models' abilities in diverse ways is therefore critical to the field. In this paper, we explore the reliability of models, where we define a reliable model as one that not only achieves strong predictive performance but also performs well consistently over many decision-making tasks involving uncertainty (e.g., selective prediction, open set recognition), robust generalization (e.g., accuracy and proper scoring rules such as log-likelihood on in- and out-of-distribution datasets), and adaptation (e.g., active learning, few-shot uncertainty). We devise 10 types of tasks over 40 datasets in order to evaluate different aspects of reliability on both vision and language domains. To improve reliability, we developed ViT-Plex and T5-Plex, pretrained large model extensions for vision and language modalities, respectively. Plex greatly improves the state-of-the-art across reliability tasks, and simplifies the traditional protocol as it improves the out-of-the-box performance and does not require designing scores or tuning the model for each task. We demonstrate scaling effects over model sizes up to 1B parameters and pretraining dataset sizes up to 4B examples. We also demonstrate Plex's capabilities on challenging tasks including zero-shot open set recognition, active learning, and uncertainty in conversational language understanding. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2207.07411v1-abstract-full').style.display = 'none'; document.getElementById('2207.07411v1-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 15 July, 2022; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> July 2022. </p> <p class="comments is-size-7"> <span class="has-text-black-bis has-text-weight-semibold">Comments:</span> <span class="has-text-grey-dark mathjax">Code available at https://goo.gle/plex-code</span> </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/2101.04526">arXiv:2101.04526</a> <span> [<a href="https://arxiv.org/pdf/2101.04526">pdf</a>, <a href="https://arxiv.org/format/2101.04526">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Machine Learning">cs.LG</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Computers and Society">cs.CY</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Information Retrieval">cs.IR</span> </div> </div> <p class="title is-5 mathjax"> Measuring Recommender System Effects with Simulated Users </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Yao%2C+S">Sirui Yao</a>, <a href="/search/cs?searchtype=author&query=Halpern%2C+Y">Yoni Halpern</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Wang%2C+X">Xuezhi Wang</a>, <a href="/search/cs?searchtype=author&query=Lee%2C+K">Kang Lee</a>, <a href="/search/cs?searchtype=author&query=Prost%2C+F">Flavien Prost</a>, <a href="/search/cs?searchtype=author&query=Chi%2C+E+H">Ed H. Chi</a>, <a href="/search/cs?searchtype=author&query=Chen%2C+J">Jilin Chen</a>, <a href="/search/cs?searchtype=author&query=Beutel%2C+A">Alex Beutel</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="2101.04526v1-abstract-short" style="display: inline;"> Imagine a food recommender system -- how would we check if it is \emph{causing} and fostering unhealthy eating habits or merely reflecting users' interests? How much of a user's experience over time with a recommender is caused by the recommender system's choices and biases, and how much is based on the user's preferences and biases? Popularity bias and filter bubbles are two of the most well-stud… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2101.04526v1-abstract-full').style.display = 'inline'; document.getElementById('2101.04526v1-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="2101.04526v1-abstract-full" style="display: none;"> Imagine a food recommender system -- how would we check if it is \emph{causing} and fostering unhealthy eating habits or merely reflecting users' interests? How much of a user's experience over time with a recommender is caused by the recommender system's choices and biases, and how much is based on the user's preferences and biases? Popularity bias and filter bubbles are two of the most well-studied recommender system biases, but most of the prior research has focused on understanding the system behavior in a single recommendation step. How do these biases interplay with user behavior, and what types of user experiences are created from repeated interactions? In this work, we offer a simulation framework for measuring the impact of a recommender system under different types of user behavior. Using this simulation framework, we can (a) isolate the effect of the recommender system from the user preferences, and (b) examine how the system performs not just on average for an "average user" but also the extreme experiences under atypical user behavior. As part of the simulation framework, we propose a set of evaluation metrics over the simulations to understand the recommender system's behavior. Finally, we present two empirical case studies -- one on traditional collaborative filtering in MovieLens and one on a large-scale production recommender system -- to understand how popularity bias manifests over time. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2101.04526v1-abstract-full').style.display = 'none'; document.getElementById('2101.04526v1-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 12 January, 2021; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> January 2021. </p> <p class="comments is-size-7"> <span class="has-text-black-bis has-text-weight-semibold">Comments:</span> <span class="has-text-grey-dark mathjax">Presented at Second Workshop on Fairness, Accountability, Transparency, Ethics and Society on the Web (FATES 2020) with the title "Beyond Next Step Bias: Trajectory Simulation for Understanding Recommender System Behavior"</span> </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/2010.07410">arXiv:2010.07410</a> <span> [<a href="https://arxiv.org/pdf/2010.07410">pdf</a>, <a href="https://arxiv.org/format/2010.07410">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Computation and Language">cs.CL</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Social and Information Networks">cs.SI</span> </div> </div> <p class="title is-5 mathjax"> Six Attributes of Unhealthy Conversation </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Price%2C+I">Ilan Price</a>, <a href="/search/cs?searchtype=author&query=Gifford-Moore%2C+J">Jordan Gifford-Moore</a>, <a href="/search/cs?searchtype=author&query=Fleming%2C+J">Jory Fleming</a>, <a href="/search/cs?searchtype=author&query=Musker%2C+S">Saul Musker</a>, <a href="/search/cs?searchtype=author&query=Roichman%2C+M">Maayan Roichman</a>, <a href="/search/cs?searchtype=author&query=Sylvain%2C+G">Guillaume Sylvain</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Dixon%2C+L">Lucas Dixon</a>, <a href="/search/cs?searchtype=author&query=Sorensen%2C+J">Jeffrey Sorensen</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="2010.07410v1-abstract-short" style="display: inline;"> We present a new dataset of approximately 44000 comments labeled by crowdworkers. Each comment is labelled as either 'healthy' or 'unhealthy', in addition to binary labels for the presence of six potentially 'unhealthy' sub-attributes: (1) hostile; (2) antagonistic, insulting, provocative or trolling; (3) dismissive; (4) condescending or patronising; (5) sarcastic; and/or (6) an unfair generalisat… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2010.07410v1-abstract-full').style.display = 'inline'; document.getElementById('2010.07410v1-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="2010.07410v1-abstract-full" style="display: none;"> We present a new dataset of approximately 44000 comments labeled by crowdworkers. Each comment is labelled as either 'healthy' or 'unhealthy', in addition to binary labels for the presence of six potentially 'unhealthy' sub-attributes: (1) hostile; (2) antagonistic, insulting, provocative or trolling; (3) dismissive; (4) condescending or patronising; (5) sarcastic; and/or (6) an unfair generalisation. Each label also has an associated confidence score. We argue that there is a need for datasets which enable research based on a broad notion of 'unhealthy online conversation'. We build this typology to encompass a substantial proportion of the individual comments which contribute to unhealthy online conversation. For some of these attributes, this is the first publicly available dataset of this scale. We explore the quality of the dataset, present some summary statistics and initial models to illustrate the utility of this data, and highlight limitations and directions for further research. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2010.07410v1-abstract-full').style.display = 'none'; document.getElementById('2010.07410v1-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 14 October, 2020; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> October 2020. </p> <p class="comments is-size-7"> <span class="has-text-black-bis has-text-weight-semibold">Comments:</span> <span class="has-text-grey-dark mathjax">Appearing in the 4th Workshop on Online Abuse and Harms (2020)</span> </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/2006.13114">arXiv:2006.13114</a> <span> [<a href="https://arxiv.org/pdf/2006.13114">pdf</a>, <a href="https://arxiv.org/format/2006.13114">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Machine Learning">cs.LG</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Machine Learning">stat.ML</span> </div> </div> <p class="title is-5 mathjax"> Fairness without Demographics through Adversarially Reweighted Learning </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Lahoti%2C+P">Preethi Lahoti</a>, <a href="/search/cs?searchtype=author&query=Beutel%2C+A">Alex Beutel</a>, <a href="/search/cs?searchtype=author&query=Chen%2C+J">Jilin Chen</a>, <a href="/search/cs?searchtype=author&query=Lee%2C+K">Kang Lee</a>, <a href="/search/cs?searchtype=author&query=Prost%2C+F">Flavien Prost</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Wang%2C+X">Xuezhi Wang</a>, <a href="/search/cs?searchtype=author&query=Chi%2C+E+H">Ed H. Chi</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="2006.13114v3-abstract-short" style="display: inline;"> Much of the previous machine learning (ML) fairness literature assumes that protected features such as race and sex are present in the dataset, and relies upon them to mitigate fairness concerns. However, in practice factors like privacy and regulation often preclude the collection of protected features, or their use for training or inference, severely limiting the applicability of traditional fai… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2006.13114v3-abstract-full').style.display = 'inline'; document.getElementById('2006.13114v3-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="2006.13114v3-abstract-full" style="display: none;"> Much of the previous machine learning (ML) fairness literature assumes that protected features such as race and sex are present in the dataset, and relies upon them to mitigate fairness concerns. However, in practice factors like privacy and regulation often preclude the collection of protected features, or their use for training or inference, severely limiting the applicability of traditional fairness research. Therefore we ask: How can we train an ML model to improve fairness when we do not even know the protected group memberships? In this work we address this problem by proposing Adversarially Reweighted Learning (ARL). In particular, we hypothesize that non-protected features and task labels are valuable for identifying fairness issues, and can be used to co-train an adversarial reweighting approach for improving fairness. Our results show that {ARL} improves Rawlsian Max-Min fairness, with notable AUC improvements for worst-case protected groups in multiple datasets, outperforming state-of-the-art alternatives. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2006.13114v3-abstract-full').style.display = 'none'; document.getElementById('2006.13114v3-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 3 November, 2020; <span class="has-text-black-bis has-text-weight-semibold">v1</span> submitted 23 June, 2020; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> June 2020. </p> <p class="comments is-size-7"> <span class="has-text-black-bis has-text-weight-semibold">Comments:</span> <span class="has-text-grey-dark mathjax">To appear at 34th Conference on Neural Information Processing Systems (NeurIPS 2020), Vancouver, Canada</span> </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/2006.00998">arXiv:2006.00998</a> <span> [<a href="https://arxiv.org/pdf/2006.00998">pdf</a>, <a href="https://arxiv.org/format/2006.00998">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Computation and Language">cs.CL</span> </div> </div> <p class="title is-5 mathjax"> Toxicity Detection: Does Context Really Matter? </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Pavlopoulos%2C+J">John Pavlopoulos</a>, <a href="/search/cs?searchtype=author&query=Sorensen%2C+J">Jeffrey Sorensen</a>, <a href="/search/cs?searchtype=author&query=Dixon%2C+L">Lucas Dixon</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Androutsopoulos%2C+I">Ion Androutsopoulos</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="2006.00998v1-abstract-short" style="display: inline;"> Moderation is crucial to promoting healthy on-line discussions. Although several `toxicity' detection datasets and models have been published, most of them ignore the context of the posts, implicitly assuming that comments maybe judged independently. We investigate this assumption by focusing on two questions: (a) does context affect the human judgement, and (b) does conditioning on context improv… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2006.00998v1-abstract-full').style.display = 'inline'; document.getElementById('2006.00998v1-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="2006.00998v1-abstract-full" style="display: none;"> Moderation is crucial to promoting healthy on-line discussions. Although several `toxicity' detection datasets and models have been published, most of them ignore the context of the posts, implicitly assuming that comments maybe judged independently. We investigate this assumption by focusing on two questions: (a) does context affect the human judgement, and (b) does conditioning on context improve performance of toxicity detection systems? We experiment with Wikipedia conversations, limiting the notion of context to the previous post in the thread and the discussion title. We find that context can both amplify or mitigate the perceived toxicity of posts. Moreover, a small but significant subset of manually labeled posts (5% in one of our experiments) end up having the opposite toxicity labels if the annotators are not provided with context. Surprisingly, we also find no evidence that context actually improves the performance of toxicity classifiers, having tried a range of classifiers and mechanisms to make them context aware. This points to the need for larger datasets of comments annotated in context. We make our code and data publicly available. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2006.00998v1-abstract-full').style.display = 'none'; document.getElementById('2006.00998v1-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 1 June, 2020; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> June 2020. </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/2004.05476">arXiv:2004.05476</a> <span> [<a href="https://arxiv.org/pdf/2004.05476">pdf</a>, <a href="https://arxiv.org/format/2004.05476">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Computation and Language">cs.CL</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Computers and Society">cs.CY</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Information Retrieval">cs.IR</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Machine Learning">cs.LG</span> </div> </div> <p class="title is-5 mathjax"> Classifying Constructive Comments </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Kolhatkar%2C+V">Varada Kolhatkar</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Sorensen%2C+J">Jeffrey Sorensen</a>, <a href="/search/cs?searchtype=author&query=Dixon%2C+L">Lucas Dixon</a>, <a href="/search/cs?searchtype=author&query=Taboada%2C+M">Maite Taboada</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="2004.05476v4-abstract-short" style="display: inline;"> We introduce the Constructive Comments Corpus (C3), comprised of 12,000 annotated news comments, intended to help build new tools for online communities to improve the quality of their discussions. We define constructive comments as high-quality comments that make a contribution to the conversation. We explain the crowd worker annotation scheme and define a taxonomy of sub-characteristics of const… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2004.05476v4-abstract-full').style.display = 'inline'; document.getElementById('2004.05476v4-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="2004.05476v4-abstract-full" style="display: none;"> We introduce the Constructive Comments Corpus (C3), comprised of 12,000 annotated news comments, intended to help build new tools for online communities to improve the quality of their discussions. We define constructive comments as high-quality comments that make a contribution to the conversation. We explain the crowd worker annotation scheme and define a taxonomy of sub-characteristics of constructiveness. The quality of the annotation scheme and the resulting dataset is evaluated using measurements of inter-annotator agreement, expert assessment of a sample, and by the constructiveness sub-characteristics, which we show provide a proxy for the general constructiveness concept. We provide models for constructiveness trained on C3 using both feature-based and a variety of deep learning approaches and demonstrate that these models capture general rather than topic- or domain-specific characteristics of constructiveness, through domain adaptation experiments. We examine the role that length plays in our models, as comment length could be easily gamed if models depend heavily upon this feature. By examining the errors made by each model and their distribution by length, we show that the best performing models are less correlated with comment length.The constructiveness corpus and our experiments pave the way for a moderation tool focused on promoting comments that make a contribution, rather than only filtering out undesirable content. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('2004.05476v4-abstract-full').style.display = 'none'; document.getElementById('2004.05476v4-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 4 August, 2020; <span class="has-text-black-bis has-text-weight-semibold">v1</span> submitted 11 April, 2020; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> April 2020. </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/1911.01916">arXiv:1911.01916</a> <span> [<a href="https://arxiv.org/pdf/1911.01916">pdf</a>, <a href="https://arxiv.org/format/1911.01916">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Machine Learning">cs.LG</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Machine Learning">stat.ML</span> </div> </div> <p class="title is-5 mathjax"> Practical Compositional Fairness: Understanding Fairness in Multi-Component Recommender Systems </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Wang%2C+X">Xuezhi Wang</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Sinha%2C+A">Anu Sinha</a>, <a href="/search/cs?searchtype=author&query=Prost%2C+F">Flavien Prost</a>, <a href="/search/cs?searchtype=author&query=Chi%2C+E+H">Ed H. Chi</a>, <a href="/search/cs?searchtype=author&query=Chen%2C+J">Jilin Chen</a>, <a href="/search/cs?searchtype=author&query=Beutel%2C+A">Alex Beutel</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="1911.01916v4-abstract-short" style="display: inline;"> How can we build recommender systems to take into account fairness? Real-world recommender systems are often composed of multiple models, built by multiple teams. However, most research on fairness focuses on improving fairness in a single model. Further, recent research on classification fairness has shown that combining multiple "fair" classifiers can still result in an "unfair" classification s… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('1911.01916v4-abstract-full').style.display = 'inline'; document.getElementById('1911.01916v4-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="1911.01916v4-abstract-full" style="display: none;"> How can we build recommender systems to take into account fairness? Real-world recommender systems are often composed of multiple models, built by multiple teams. However, most research on fairness focuses on improving fairness in a single model. Further, recent research on classification fairness has shown that combining multiple "fair" classifiers can still result in an "unfair" classification system. This presents a significant challenge: how do we understand and improve fairness in recommender systems composed of multiple components? In this paper, we study the compositionality of recommender fairness. We consider two recently proposed fairness ranking metrics: equality of exposure and pairwise ranking accuracy. While we show that fairness in recommendation is not guaranteed to compose, we provide theory for a set of conditions under which fairness of individual models does compose. We then present an analytical framework for both understanding whether a real system's signals can achieve compositional fairness, and improving which component would have the greatest impact on the fairness of the overall system. In addition to the theoretical results, we find on multiple datasets -- including a large-scale real-world recommender system -- that the overall system's end-to-end fairness is largely achievable by improving fairness in individual components. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('1911.01916v4-abstract-full').style.display = 'none'; document.getElementById('1911.01916v4-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 25 January, 2021; <span class="has-text-black-bis has-text-weight-semibold">v1</span> submitted 5 November, 2019; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> November 2019. </p> <p class="comments is-size-7"> <span class="has-text-black-bis has-text-weight-semibold">Comments:</span> <span class="has-text-grey-dark mathjax">WSDM 2021</span> </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/1908.02810">arXiv:1908.02810</a> <span> [<a href="https://arxiv.org/pdf/1908.02810">pdf</a>, <a href="https://arxiv.org/format/1908.02810">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Machine Learning">cs.LG</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Computation and Language">cs.CL</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Machine Learning">stat.ML</span> </div> </div> <p class="title is-5 mathjax"> Debiasing Embeddings for Reduced Gender Bias in Text Classification </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Prost%2C+F">Flavien Prost</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Bolukbasi%2C+T">Tolga Bolukbasi</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="1908.02810v1-abstract-short" style="display: inline;"> (Bolukbasi et al., 2016) demonstrated that pretrained word embeddings can inherit gender bias from the data they were trained on. We investigate how this bias affects downstream classification tasks, using the case study of occupation classification (De-Arteaga et al.,2019). We show that traditional techniques for debiasing embeddings can actually worsen the bias of the downstream classifier by pr… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('1908.02810v1-abstract-full').style.display = 'inline'; document.getElementById('1908.02810v1-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="1908.02810v1-abstract-full" style="display: none;"> (Bolukbasi et al., 2016) demonstrated that pretrained word embeddings can inherit gender bias from the data they were trained on. We investigate how this bias affects downstream classification tasks, using the case study of occupation classification (De-Arteaga et al.,2019). We show that traditional techniques for debiasing embeddings can actually worsen the bias of the downstream classifier by providing a less noisy channel for communicating gender information. With a relatively minor adjustment, however, we show how these same techniques can be used to simultaneously reduce bias and maintain high classification accuracy. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('1908.02810v1-abstract-full').style.display = 'none'; document.getElementById('1908.02810v1-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 7 August, 2019; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> August 2019. </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/1903.04561">arXiv:1903.04561</a> <span> [<a href="https://arxiv.org/pdf/1903.04561">pdf</a>, <a href="https://arxiv.org/format/1903.04561">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Machine Learning">cs.LG</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Computation and Language">cs.CL</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Machine Learning">stat.ML</span> </div> </div> <p class="title is-5 mathjax"> Nuanced Metrics for Measuring Unintended Bias with Real Data for Text Classification </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Borkan%2C+D">Daniel Borkan</a>, <a href="/search/cs?searchtype=author&query=Dixon%2C+L">Lucas Dixon</a>, <a href="/search/cs?searchtype=author&query=Sorensen%2C+J">Jeffrey Sorensen</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Vasserman%2C+L">Lucy Vasserman</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="1903.04561v2-abstract-short" style="display: inline;"> Unintended bias in Machine Learning can manifest as systemic differences in performance for different demographic groups, potentially compounding existing challenges to fairness in society at large. In this paper, we introduce a suite of threshold-agnostic metrics that provide a nuanced view of this unintended bias, by considering the various ways that a classifier's score distribution can vary ac… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('1903.04561v2-abstract-full').style.display = 'inline'; document.getElementById('1903.04561v2-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="1903.04561v2-abstract-full" style="display: none;"> Unintended bias in Machine Learning can manifest as systemic differences in performance for different demographic groups, potentially compounding existing challenges to fairness in society at large. In this paper, we introduce a suite of threshold-agnostic metrics that provide a nuanced view of this unintended bias, by considering the various ways that a classifier's score distribution can vary across designated groups. We also introduce a large new test set of online comments with crowd-sourced annotations for identity references. We use this to show how our metrics can be used to find new and potentially subtle unintended bias in existing public models. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('1903.04561v2-abstract-full').style.display = 'none'; document.getElementById('1903.04561v2-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 8 May, 2019; <span class="has-text-black-bis has-text-weight-semibold">v1</span> submitted 11 March, 2019; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> March 2019. </p> <p class="comments is-size-7"> <span class="has-text-black-bis has-text-weight-semibold">Comments:</span> <span class="has-text-grey-dark mathjax">Updated to fix typo in Equation 4</span> </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/1903.02088">arXiv:1903.02088</a> <span> [<a href="https://arxiv.org/pdf/1903.02088">pdf</a>, <a href="https://arxiv.org/format/1903.02088">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Machine Learning">stat.ML</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Machine Learning">cs.LG</span> </div> </div> <p class="title is-5 mathjax"> Limitations of Pinned AUC for Measuring Unintended Bias </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Borkan%2C+D">Daniel Borkan</a>, <a href="/search/cs?searchtype=author&query=Dixon%2C+L">Lucas Dixon</a>, <a href="/search/cs?searchtype=author&query=Li%2C+J">John Li</a>, <a href="/search/cs?searchtype=author&query=Sorensen%2C+J">Jeffrey Sorensen</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Vasserman%2C+L">Lucy Vasserman</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="1903.02088v1-abstract-short" style="display: inline;"> This report examines the Pinned AUC metric introduced and highlights some of its limitations. Pinned AUC provides a threshold-agnostic measure of unintended bias in a classification model, inspired by the ROC-AUC metric. However, as we highlight in this report, there are ways that the metric can obscure different kinds of unintended biases when the underlying class distributions on which bias is b… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('1903.02088v1-abstract-full').style.display = 'inline'; document.getElementById('1903.02088v1-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="1903.02088v1-abstract-full" style="display: none;"> This report examines the Pinned AUC metric introduced and highlights some of its limitations. Pinned AUC provides a threshold-agnostic measure of unintended bias in a classification model, inspired by the ROC-AUC metric. However, as we highlight in this report, there are ways that the metric can obscure different kinds of unintended biases when the underlying class distributions on which bias is being measured are not carefully controlled. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('1903.02088v1-abstract-full').style.display = 'none'; document.getElementById('1903.02088v1-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 5 March, 2019; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> March 2019. </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/1810.13181">arXiv:1810.13181</a> <span> [<a href="https://arxiv.org/pdf/1810.13181">pdf</a>, <a href="https://arxiv.org/format/1810.13181">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Computation and Language">cs.CL</span> </div> </div> <p class="title is-5 mathjax"> WikiConv: A Corpus of the Complete Conversational History of a Large Online Collaborative Community </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Hua%2C+Y">Yiqing Hua</a>, <a href="/search/cs?searchtype=author&query=Danescu-Niculescu-Mizil%2C+C">Cristian Danescu-Niculescu-Mizil</a>, <a href="/search/cs?searchtype=author&query=Taraborelli%2C+D">Dario Taraborelli</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Sorensen%2C+J">Jeffery Sorensen</a>, <a href="/search/cs?searchtype=author&query=Dixon%2C+L">Lucas Dixon</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="1810.13181v1-abstract-short" style="display: inline;"> We present a corpus that encompasses the complete history of conversations between contributors to Wikipedia, one of the largest online collaborative communities. By recording the intermediate states of conversations---including not only comments and replies, but also their modifications, deletions and restorations---this data offers an unprecedented view of online conversation. This level of deta… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('1810.13181v1-abstract-full').style.display = 'inline'; document.getElementById('1810.13181v1-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="1810.13181v1-abstract-full" style="display: none;"> We present a corpus that encompasses the complete history of conversations between contributors to Wikipedia, one of the largest online collaborative communities. By recording the intermediate states of conversations---including not only comments and replies, but also their modifications, deletions and restorations---this data offers an unprecedented view of online conversation. This level of detail supports new research questions pertaining to the process (and challenges) of large-scale online collaboration. We illustrate the corpus' potential with two case studies that highlight new perspectives on earlier work. First, we explore how a person's conversational behavior depends on how they relate to the discussion's venue. Second, we show that community moderation of toxic behavior happens at a higher rate than previously estimated. Finally the reconstruction framework is designed to be language agnostic, and we show that it can extract high quality conversational data in both Chinese and English. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('1810.13181v1-abstract-full').style.display = 'none'; document.getElementById('1810.13181v1-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 31 October, 2018; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> October 2018. </p> <p class="comments is-size-7"> <span class="has-text-black-bis has-text-weight-semibold">Journal ref:</span> Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/1805.05345">arXiv:1805.05345</a> <span> [<a href="https://arxiv.org/pdf/1805.05345">pdf</a>, <a href="https://arxiv.org/format/1805.05345">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Computation and Language">cs.CL</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Artificial Intelligence">cs.AI</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Computers and Society">cs.CY</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Human-Computer Interaction">cs.HC</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Physics and Society">physics.soc-ph</span> </div> </div> <p class="title is-5 mathjax"> Conversations Gone Awry: Detecting Early Signs of Conversational Failure </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Zhang%2C+J">Justine Zhang</a>, <a href="/search/cs?searchtype=author&query=Chang%2C+J+P">Jonathan P. Chang</a>, <a href="/search/cs?searchtype=author&query=Danescu-Niculescu-Mizil%2C+C">Cristian Danescu-Niculescu-Mizil</a>, <a href="/search/cs?searchtype=author&query=Dixon%2C+L">Lucas Dixon</a>, <a href="/search/cs?searchtype=author&query=Hua%2C+Y">Yiqing Hua</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Taraborelli%2C+D">Dario Taraborelli</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="1805.05345v1-abstract-short" style="display: inline;"> One of the main challenges online social systems face is the prevalence of antisocial behavior, such as harassment and personal attacks. In this work, we introduce the task of predicting from the very start of a conversation whether it will get out of hand. As opposed to detecting undesirable behavior after the fact, this task aims to enable early, actionable prediction at a time when the conversa… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('1805.05345v1-abstract-full').style.display = 'inline'; document.getElementById('1805.05345v1-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="1805.05345v1-abstract-full" style="display: none;"> One of the main challenges online social systems face is the prevalence of antisocial behavior, such as harassment and personal attacks. In this work, we introduce the task of predicting from the very start of a conversation whether it will get out of hand. As opposed to detecting undesirable behavior after the fact, this task aims to enable early, actionable prediction at a time when the conversation might still be salvaged. To this end, we develop a framework for capturing pragmatic devices---such as politeness strategies and rhetorical prompts---used to start a conversation, and analyze their relation to its future trajectory. Applying this framework in a controlled setting, we demonstrate the feasibility of detecting early warning signs of antisocial behavior in online discussions. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('1805.05345v1-abstract-full').style.display = 'none'; document.getElementById('1805.05345v1-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 14 May, 2018; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> May 2018. </p> <p class="comments is-size-7"> <span class="has-text-black-bis has-text-weight-semibold">Comments:</span> <span class="has-text-grey-dark mathjax">To appear in the Proceedings of ACL 2018, 15 pages, 1 figure. Data, quiz, code and additional information at http://www.cs.cornell.edu/~cristian/Conversations_gone_awry.html</span> </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/1610.08914">arXiv:1610.08914</a> <span> [<a href="https://arxiv.org/pdf/1610.08914">pdf</a>, <a href="https://arxiv.org/format/1610.08914">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Computation and Language">cs.CL</span> </div> </div> <p class="title is-5 mathjax"> Ex Machina: Personal Attacks Seen at Scale </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Wulczyn%2C+E">Ellery Wulczyn</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Dixon%2C+L">Lucas Dixon</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="1610.08914v2-abstract-short" style="display: inline;"> The damage personal attacks cause to online discourse motivates many platforms to try to curb the phenomenon. However, understanding the prevalence and impact of personal attacks in online platforms at scale remains surprisingly difficult. The contribution of this paper is to develop and illustrate a method that combines crowdsourcing and machine learning to analyze personal attacks at scale. We s… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('1610.08914v2-abstract-full').style.display = 'inline'; document.getElementById('1610.08914v2-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="1610.08914v2-abstract-full" style="display: none;"> The damage personal attacks cause to online discourse motivates many platforms to try to curb the phenomenon. However, understanding the prevalence and impact of personal attacks in online platforms at scale remains surprisingly difficult. The contribution of this paper is to develop and illustrate a method that combines crowdsourcing and machine learning to analyze personal attacks at scale. We show an evaluation method for a classifier in terms of the aggregated number of crowd-workers it can approximate. We apply our methodology to English Wikipedia, generating a corpus of over 100k high quality human-labeled comments and 63M machine-labeled ones from a classifier that is as good as the aggregate of 3 crowd-workers, as measured by the area under the ROC curve and Spearman correlation. Using this corpus of machine-labeled scores, our methodology allows us to explore some of the open questions about the nature of online personal attacks. This reveals that the majority of personal attacks on Wikipedia are not the result of a few malicious users, nor primarily the consequence of allowing anonymous contributions from unregistered users. <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('1610.08914v2-abstract-full').style.display = 'none'; document.getElementById('1610.08914v2-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 25 February, 2017; <span class="has-text-black-bis has-text-weight-semibold">v1</span> submitted 27 October, 2016; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> October 2016. </p> </li> <li class="arxiv-result"> <div class="is-marginless"> <p class="list-title is-inline-block"><a href="https://arxiv.org/abs/1202.4134">arXiv:1202.4134</a> <span> [<a href="https://arxiv.org/pdf/1202.4134">pdf</a>, <a href="https://arxiv.org/format/1202.4134">other</a>] </span> </p> <div class="tags is-inline-block"> <span class="tag is-small is-link tooltip is-tooltip-top" data-tooltip="Computer Science and Game Theory">cs.GT</span> <span class="tag is-small is-grey tooltip is-tooltip-top" data-tooltip="Data Structures and Algorithms">cs.DS</span> </div> </div> <p class="title is-5 mathjax"> On the Implications of Lookahead Search in Game Playing </p> <p class="authors"> <span class="search-hit">Authors:</span> <a href="/search/cs?searchtype=author&query=Mirrokni%2C+V">Vahab Mirrokni</a>, <a href="/search/cs?searchtype=author&query=Thain%2C+N">Nithum Thain</a>, <a href="/search/cs?searchtype=author&query=Vetta%2C+A">Adrian Vetta</a> </p> <p class="abstract mathjax"> <span class="has-text-black-bis has-text-weight-semibold">Abstract</span>: <span class="abstract-short has-text-grey-dark mathjax" id="1202.4134v1-abstract-short" style="display: inline;"> Lookahead search is perhaps the most natural and widely used game playing strategy. Given the practical importance of the method, the aim of this paper is to provide a theoretical performance examination of lookahead search in a wide variety of applications. To determine a strategy play using lookahead search}, each agent predicts multiple levels of possible re-actions to her move (via the use o… <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('1202.4134v1-abstract-full').style.display = 'inline'; document.getElementById('1202.4134v1-abstract-short').style.display = 'none';">▽ More</a> </span> <span class="abstract-full has-text-grey-dark mathjax" id="1202.4134v1-abstract-full" style="display: none;"> Lookahead search is perhaps the most natural and widely used game playing strategy. Given the practical importance of the method, the aim of this paper is to provide a theoretical performance examination of lookahead search in a wide variety of applications. To determine a strategy play using lookahead search}, each agent predicts multiple levels of possible re-actions to her move (via the use of a search tree), and then chooses the play that optimizes her future payoff accounting for these re-actions. There are several choices of optimization function the agents can choose, where the most appropriate choice of function will depend on the specifics of the actual game - we illustrate this in our examples. Furthermore, the type of search tree chosen by computationally-constrained agent can vary. We focus on the case where agents can evaluate only a bounded number, $k$, of moves into the future. That is, we use depth $k$ search trees and call this approach {\em k-lookahead search}. We apply our method in five well-known settings: AdWord auctions; industrial organization (Cournot's model); congestion games; valid-utility games and basic-utility games; cost-sharing network design games. We consider two questions. First, what is the expected social quality of outcome when agents apply lookahead search? Second, what interactive behaviours can be exhibited when players use lookahead search? <a class="is-size-7" style="white-space: nowrap;" onclick="document.getElementById('1202.4134v1-abstract-full').style.display = 'none'; document.getElementById('1202.4134v1-abstract-short').style.display = 'inline';">△ Less</a> </span> </p> <p class="is-size-7"><span class="has-text-black-bis has-text-weight-semibold">Submitted</span> 19 February, 2012; <span class="has-text-black-bis has-text-weight-semibold">originally announced</span> February 2012. </p> </li> </ol> <div class="is-hidden-tablet">  <span class="help" style="display: inline-block;"><a href="https://github.com/arXiv/arxiv-search/releases">Search v0.5.6 released 2020-02-24</a>  </span> </div> </div> </main> <footer> <div class="columns is-desktop" role="navigation" aria-label="Secondary">  <div class="column"> <div class="columns"> <div class="column"> <ul class="nav-spaced"> <li><a href="https://info.arxiv.org/about">About</a></li> <li><a href="https://info.arxiv.org/help">Help</a></li> </ul> </div> <div class="column"> <ul class="nav-spaced"> <li> <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 512 512" class="icon filter-black" role="presentation"><title>contact arXiv</title><desc>Click here to contact arXiv</desc><path d="M502.3 190.8c3.9-3.1 9.7-.2 9.7 4.7V400c0 26.5-21.5 48-48 48H48c-26.5 0-48-21.5-48-48V195.6c0-5 5.7-7.8 9.7-4.7 22.4 17.4 52.1 39.5 154.1 113.6 21.1 15.4 56.7 47.8 92.2 47.6 35.7.3 72-32.8 92.3-47.6 102-74.1 131.6-96.3 154-113.7zM256 320c23.2.4 56.6-29.2 73.4-41.4 132.7-96.3 142.8-104.7 173.4-128.7 5.8-4.5 9.2-11.5 9.2-18.9v-19c0-26.5-21.5-48-48-48H48C21.5 64 0 85.5 0 112v19c0 7.4 3.4 14.3 9.2 18.9 30.6 23.9 40.7 32.4 173.4 128.7 16.8 12.2 50.2 41.8 73.4 41.4z"/></svg> <a href="https://info.arxiv.org/help/contact.html"> Contact</a> </li> <li> <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 512 512" class="icon filter-black" role="presentation"><title>subscribe to arXiv mailings</title><desc>Click here to subscribe</desc><path d="M476 3.2L12.5 270.6c-18.1 10.4-15.8 35.6 2.2 43.2L121 358.4l287.3-253.2c5.5-4.9 13.3 2.6 8.6 8.3L176 407v80.5c0 23.6 28.5 32.9 42.5 15.8L282 426l124.6 52.2c14.2 6 30.4-2.9 33-18.2l72-432C515 7.8 493.3-6.8 476 3.2z"/></svg> <a href="https://info.arxiv.org/help/subscribe"> Subscribe</a> </li> </ul> </div> </div> </div>   <div class="column"> <div class="columns"> <div class="column"> <ul class="nav-spaced"> <li><a href="https://info.arxiv.org/help/license/index.html">Copyright</a></li> <li><a href="https://info.arxiv.org/help/policies/privacy_policy.html">Privacy Policy</a></li> </ul> </div> <div class="column sorry-app-links"> <ul class="nav-spaced"> <li><a href="https://info.arxiv.org/help/web_accessibility.html">Web Accessibility Assistance</a></li> <li> <p class="help"> <a class="a11y-main-link" href="https://status.arxiv.org" target="_blank">arXiv Operational Status <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 256 512" class="icon filter-dark_grey" role="presentation"><path d="M224.3 273l-136 136c-9.4 9.4-24.6 9.4-33.9 0l-22.6-22.6c-9.4-9.4-9.4-24.6 0-33.9l96.4-96.4-96.4-96.4c-9.4-9.4-9.4-24.6 0-33.9L54.3 103c9.4-9.4 24.6-9.4 33.9 0l136 136c9.5 9.4 9.5 24.6.1 34z"/></svg></a><br> Get status notifications via <a class="is-link" href="https://subscribe.sorryapp.com/24846f03/email/new" target="_blank"><svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 512 512" class="icon filter-black" role="presentation"><path d="M502.3 190.8c3.9-3.1 9.7-.2 9.7 4.7V400c0 26.5-21.5 48-48 48H48c-26.5 0-48-21.5-48-48V195.6c0-5 5.7-7.8 9.7-4.7 22.4 17.4 52.1 39.5 154.1 113.6 21.1 15.4 56.7 47.8 92.2 47.6 35.7.3 72-32.8 92.3-47.6 102-74.1 131.6-96.3 154-113.7zM256 320c23.2.4 56.6-29.2 73.4-41.4 132.7-96.3 142.8-104.7 173.4-128.7 5.8-4.5 9.2-11.5 9.2-18.9v-19c0-26.5-21.5-48-48-48H48C21.5 64 0 85.5 0 112v19c0 7.4 3.4 14.3 9.2 18.9 30.6 23.9 40.7 32.4 173.4 128.7 16.8 12.2 50.2 41.8 73.4 41.4z"/></svg>email</a> or <a class="is-link" href="https://subscribe.sorryapp.com/24846f03/slack/new" target="_blank"><svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 448 512" class="icon filter-black" role="presentation"><path d="M94.12 315.1c0 25.9-21.16 47.06-47.06 47.06S0 341 0 315.1c0-25.9 21.16-47.06 47.06-47.06h47.06v47.06zm23.72 0c0-25.9 21.16-47.06 47.06-47.06s47.06 21.16 47.06 47.06v117.84c0 25.9-21.16 47.06-47.06 47.06s-47.06-21.16-47.06-47.06V315.1zm47.06-188.98c-25.9 0-47.06-21.16-47.06-47.06S139 32 164.9 32s47.06 21.16 47.06 47.06v47.06H164.9zm0 23.72c25.9 0 47.06 21.16 47.06 47.06s-21.16 47.06-47.06 47.06H47.06C21.16 243.96 0 222.8 0 196.9s21.16-47.06 47.06-47.06H164.9zm188.98 47.06c0-25.9 21.16-47.06 47.06-47.06 25.9 0 47.06 21.16 47.06 47.06s-21.16 47.06-47.06 47.06h-47.06V196.9zm-23.72 0c0 25.9-21.16 47.06-47.06 47.06-25.9 0-47.06-21.16-47.06-47.06V79.06c0-25.9 21.16-47.06 47.06-47.06 25.9 0 47.06 21.16 47.06 47.06V196.9zM283.1 385.88c25.9 0 47.06 21.16 47.06 47.06 0 25.9-21.16 47.06-47.06 47.06-25.9 0-47.06-21.16-47.06-47.06v-47.06h47.06zm0-23.72c-25.9 0-47.06-21.16-47.06-47.06 0-25.9 21.16-47.06 47.06-47.06h117.84c25.9 0 47.06 21.16 47.06 47.06 0 25.9-21.16 47.06-47.06 47.06H283.1z"/></svg>slack</a> </p> </li> </ul> </div> </div> </div>  </div> </footer> <script src="https://static.arxiv.org/static/base/1.0.0a5/js/member_acknowledgement.js"></script> </body> </html>

CINXE.COM

Search | arXiv e-print repository