<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Johnnatan Messias on Medium]]></title>
        <description><![CDATA[Stories by Johnnatan Messias on Medium]]></description>
        <link>https://medium.com/@johnnatanmessias?source=rss-36b773245cf------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*cF-HK-UFPb7m5XOH.</url>
            <title>Stories by Johnnatan Messias on Medium</title>
            <link>https://medium.com/@johnnatanmessias?source=rss-36b773245cf------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 23:16:31 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@johnnatanmessias/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Book: The Victorian Internet]]></title>
            <link>https://medium.com/@johnnatanmessias/book-the-victorian-internet-d89927985843?source=rss-36b773245cf------2</link>
            <guid isPermaLink="false">https://medium.com/p/d89927985843</guid>
            <category><![CDATA[book-review]]></category>
            <category><![CDATA[telegraph]]></category>
            <category><![CDATA[book-recommendations]]></category>
            <category><![CDATA[tech]]></category>
            <category><![CDATA[internet]]></category>
            <dc:creator><![CDATA[Johnnatan Messias]]></dc:creator>
            <pubDate>Mon, 03 Dec 2018 22:07:25 GMT</pubDate>
            <atom:updated>2018-12-04T06:43:48.879Z</atom:updated>
            <content:encoded><![CDATA[<p>Today my recommendation is the book &quot;The Victorian Internet&quot; by Tom Standage. This book presents the story of the telegraph since its creation. Centuries ago, people were not able to communicate with others as we do nowadays.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/364/1*WNLOVl_I1_HXS6e951swAw.jpeg" /></figure><p>The art of easily sending messages to our family and friends would take weeks or even months until the message reached them. The main way of communication was based on people, horses, and, the fastest way, trains. Today it is hard to imagine ourselves without the Internet, isn’t it?</p><p>The inventor of the first improved way of communication was Claude Chappe who created a rudimentary visual-communication apparatus called the Optical Telegraph. Basically, he proposed the construction of some towers to support this apparatus on the top of the tower. Thus, people from other towers could see the symbols of the previous one and forward the message to the next one until it reached its destination. As we may note, it is not an efficient way of communication because: (i) it is not possible to send messages across far distances; (ii) it does not work during the night; and (iii) people still need to wait a long time to receive a single message from other countries due to the long distance.</p><p>To fix those issues, people started to pursue a new improvement for the Optical Telegraph. The Electrical Telegraph. This new technology could send messages over long distances and also across the ocean, joining continents to the network. For the first time, people could then send messages to their loved ones almost instantly. From this point on, newspaper around the world started to talk about this new improvement and how it could make humans borderless.</p><p>After the telegraph the world started to continue the improvements until inventors created the Internet: our beloved and great way of communication that we use for everything today. Nowadays we can send messages, videos, make calls, access tons of information and so many things we could not imagine in the past. The Internet has made everyone connected.</p><p>The interesting thing about this book is that during each of my readings, I realized that the Internet is basically the Telegraph (of course in a real rudimentary form). As with the telegraph, the principle way we exchange information on the Internet is via wires and, due to the digital computers, we use the binary code. With telegraphs, we also used wires and a code which is called Morse Code. The main way we exchange information on the Internet across oceans is also via wires and the same was for the telegraph. The book also explains how technology keeps improving over time and the most impressive thing is that for every new change and improvement, new challenges and issues are also created. It is a cycle. Also, will our descendants read a book explaining the origin of computers, social networks, and so many things that are normal for us today? Will the future be so different? What is your guess?</p><p>For much more information and stories behind the invention of the Telegraph until the creation of the Internet, I suggest the reading of this amazing book :)</p><h3>About me</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/1*vKxMKObPxWG27V_mQXqSVg.jpeg" /><figcaption><a href="http://johnnatan.me/"><strong>Johnnatan Messias</strong></a> — Ph.D Student at MPI-SWS</figcaption></figure><p>Currently, I am a Ph.D. student at the Max Planck Institute for Software Systems (<a href="https://www.mpi-sws.org/">MPI-SWS</a>) in the <a href="http://socialnetworks.mpi-sws.org/">Social Computing Research</a> group.</p><p>I received a Master degree in Computer Science from Universidade Federal de Minas Gerais (<a href="http://www.ufmg.br/">UFMG</a>). During my MsC, I also held research intern position at Max Planck Institute for Software Systems (<a href="http://www.mpi-sws.org/">MPI-SWS</a>) in Saarbrücken/Germany. I studied Computer Science at Eötvös Loránd University (<a href="https://www.elte.hu/en/">ELTE</a>) in Budapest/Hungary in 2013/2014 during my exchange program (Science Without Borders — CAPES). I received a bachelor degree in Computer Science from Universidade Federal de Ouro Preto (<a href="http://www.ufop.br/">UFOP</a>), in 2015.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d89927985843" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Demographic Data in Social Media: Why is it important?]]></title>
            <link>https://medium.com/@johnnatanmessias/demographic-data-in-social-media-why-is-it-important-ca52790dfecf?source=rss-36b773245cf------2</link>
            <guid isPermaLink="false">https://medium.com/p/ca52790dfecf</guid>
            <category><![CDATA[inequality]]></category>
            <category><![CDATA[social-media]]></category>
            <category><![CDATA[fairness]]></category>
            <category><![CDATA[twitter]]></category>
            <category><![CDATA[demographics]]></category>
            <dc:creator><![CDATA[Johnnatan Messias]]></dc:creator>
            <pubDate>Sun, 25 Feb 2018 16:04:24 GMT</pubDate>
            <atom:updated>2018-02-25T16:04:24.356Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*EZqpMZp0BDDXrLbX." /><figcaption>“A hand holding a vintage video camera with stickers on it” by <a href="https://unsplash.com/@stickermule?utm_source=medium&amp;utm_medium=referral">Sticker Mule</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Social media is considered a democratic space in which people connect and interact with each other regardless of their gender, race, or any other demographic aspect. Despite numerous efforts that explore demographic aspects in social media, it is still unclear whether social media perpetuates old inequalities from the offline world.</p><p>In this article, I will discuss the importance of demographic data in social media studies. Please, feel free to post comments about what you think regarding this article. Suggestions are always wellcome :)</p><p>For a more and detailed content please, feel free to check out my published work (<a href="http://johnnatan.me/">Link</a>) and my Master thesis (in English) available <a href="http://johnnatan.me/pdf/master-thesis_johnnatan_messias.pdf">here</a> [22].</p><h3>Introduction</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*5i6E1xH2jeW3l4rM." /><figcaption>“Brightly colored urban graffiti mural artwork on wall with equal text and tribal faces” by <a href="https://unsplash.com/@o_j_cole?utm_source=medium&amp;utm_medium=referral">Oliver Cole</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Nowadays, millions of people constantly use online social networking sites, such as Facebook and Twitter. In the third quarter of 2016, Facebook and Twitter had 1.79 billion [1] and 317 million [2] monthly active users, respectively, sharing content about their daily lives and things that happen around them. These systems have revolutionized the way we communicate, by organizing our offline social relationships in a digital form.</p><p>Increasingly, our society has been treating actions in the online and offline space in an indistinguishable way. It is now common to see cases in which content posted on Facebook or Twitter is used to vet job candidates, support divorce litigation, or terminate employees [3, 4]. Although these platforms provide a democratic space for conversations and debates, they also open space for key unsolved issues of our society, such as gender and race inequalities.</p><p>The massive popularity of online social media also provides the opportunity to detect useful characteristics and patterns about users and their interconnections. For instance, patterns are valuable for marketing and advertisement companies which capture users behavior and needs in order to promote products, specifically on a target group. In terms of groups, demographics constitutes a significant factor to cluster people and understand their behavior.</p><p>Twitter is a micro-blogging platform in which the main form of communication and interaction is based on posting text messages. It is also a very large system with lots of users. Consequently, it gains the attention of companies and researchers to explore who are these users and what they post on this social network. Although it provides a log of a number of human interaction and social connections, its data is unstructured and lacks information about demographics of users, such as gender, race, and age. So, the existing efforts that attempt to explore demographics of users on this space rely on different inference strategies. These efforts include [5, 6, 7, 8], and [9].</p><p>Despite the magnitude of Twitter and the amount of the provided information, inferring such demographic aspects of users is challenging and the existing works have several limitations due to the difficulty to identify and gather such demographic aspects. However, the potential to measure demographic aspects is immense and valuable for different research purposes and applications, such as design systems, to understand the social aspects of our society, and how different demographic groups interact with each other. This post focuses on this context. In the next section, we present the main motivations about why we really should study demographics in Social Media.</p><h3>Why is Demographic Data important?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*nkLjwPralnK4TloJ." /><figcaption>Photo by <a href="https://unsplash.com/@ibydesigns?utm_source=medium&amp;utm_medium=referral">Alex Iby</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Exploring demographic aspects is important both from the perspective of systems and from the sociological point of view. In the first, demographic aspects allow us to understand groups of users that perform certain actions and interactions within a system. As from the sociological point of view, demographic aspects also help us to understand the gender and race differences in online interactions and, consequently, to better understand our society [20].</p><p>Demographic aspects also allow us to study the linguistic differences as well as diversity in topics of interest between gender and race [21]. I believe that studying the differences in how demographic groups write content about specific topics would help systems developers to provide more transparency to users in their systems. The ability of understanding transparency is for algorithms that operate on large-scale crowdsourced data, the goal is to make the outputs of the algorithms (and the algorithm itself) transparent, as it is also important to understand the non-uniformities in the inputs to the algorithms.</p><p>Recent efforts show gender disparities on online systems which highlight a concern regarding the inequalities and the main barriers that keep a given demographic from rising beyond a certain level in a hierarchy. For example, women are under-represented in the content production in <a href="https://www.wikipedia.org/">Wikipedia</a>. Further, black tenants have much fewer chances of renting a place on <a href="https://www.airbnb.com/">Airbnb</a>. Some drivers for both <a href="https://www.uber.com/">UberX</a> and <a href="https://www.lyft.com/">Lyft</a> discriminate on the basis of the perceived race of the traveler. There is also evidence where disadvantages against female users are found among the users with high visibility [9], similar to the barrier females face in attaining higher positions in companies (“glass ceiling” effect) [10]. Therefore, identifying inequalities and asymmetries in demographics in the online world is crucial for the development of features that can promote equality and diversity in these systems.</p><p>Despite the recent surge of interest in demographic aspects of users from social networks, inferring demographic aspects is challenging and limited due to the difficult to identify and gather such aspects from the users in these systems. The main challenge is that demographic information is not directly available.</p><h3>Demographic in Social Media</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*QtRKAWsOl1eKyY29." /><figcaption>“An phone with the Facebook app open next to Scrabble pieces arranged in the words “social media”” by <a href="https://unsplash.com/@firmbee?utm_source=medium&amp;utm_medium=referral">William Iven</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>There are several studies that retrieve information from social media to analyze and predict real-world phenomena such as stock market [11], migration [12], election [13, 14], and also political leaning inference of Twitter users [15, 16].</p><p>One of the first efforts to extract and analyze demographic information presents a comparative study between the demographic distribution of gender/race of Twitter users and U.S. population [8].<br>After that, several efforts have arisen that investigate demographic information, in various social media, using different strategies for distinct purposes [5, 6, 17, 18, 19].</p><p>Recent studies focused on demographics [5, 6, 7, 9, 21] present methodologies to extract the necessary data through analysis and pattern matching of screen/full name as well as descriptions of user profiles and image in the profile status. Particularly, Chen <em>et al.</em> [23] focus on demographic inference using profile self-descriptions and profile images. They analyze five signals in order to categorize users to their demographic status. They use users’ names, self-descriptions, tweets, social networks, and profile images to infer demographic aspects (ethnicity, gender, age). Culotta <em>et al.</em> [24] describe an alternative methodology assuming that the demographic profiles of visitors to a website are correlated with the demographic profiles of followers of that website on social network (e.g. Twitter), and propose a regression model to determine demographic aspects (gender, age, ethnicity, education, income, and child status). In another study, Reis <em>et al.</em> [25] study demographic through news sharing in U.S. perspective and show that male and white users tend to be more active in terms of sharing news, biasing news audience to the interest of these demographic groups.</p><p>Another study, An <em>et al.</em> [26] examine the correlation of hashtags that are used in different demographic groups, using Face++ and users’ profile pictures. They show that this strategy is reliable and provides accurate demographic information for gender and race. More recently, Chakraborty <em>et al.</em> [27] analyze the demographic of Twitter trending topics, using Face++ as well. They use an extensive dataset collected from Twitter in order to make the first attempt to quantify and explore the demographic biases in crowdsourced recommendations. They find that a large fraction of trends is promoted by crowds whose demographics are significantly different from the overall Twitter population. More worryingly, certain demographic groups are systematically under-represented among the promoters of the trending topics.</p><h3>Inequality in Twitter Visibility</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*JOSqm4BBhNeEhhcW." /><figcaption>Photo by <a href="https://unsplash.com/@glencarrie?utm_source=medium&amp;utm_medium=referral">Glen Carrie</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Homophily states that similarity breeds social connections [28]. In our society, people tend to characterize and categorize others in terms of demographic aspects, such as gender and race [29], making them intuitively natural concepts through which people might account similarities. The existing literature is still limited in explaining the inequalities that arise from the establishments of these social connections in Twitter.</p><p>Most of the existing efforts in this space attempt to quantify and understand other factors that are highly correlated with visibility (i.e. number of followers). Particularly, Morales <em>et al.</em> [30] show that user’s reputation and her impact on society constitute significant factors of the online visibility growth. Other efforts show that the attention that users gain is correlated with his/her online behavior in terms of the frequency of their interactions [31, 32, 33]. Freitas <em>et al.</em> [34] created 120 socialbots in Twitter, with different behaviors and demographics aspects. They show that users who post more, post about specific topics, reply, favorite or mention others frequently, may acquire more followers, and consequently, more visibility. <br>Despite the extreme importance of these efforts, they do not investigate the correlation of demographic issues with visibility. Thus, our effort is complementary to theirs.</p><p>There have been efforts quantifying differences and inequalities in many other social media, including Wikipedia [35, 36] and Pinterest [37]. <br>However, the study that is closest related to ours explores gender inequalities in Twitter [9]. Authors show that gender may allow inequality to persist in terms of online visibility. This dissertation further explores race as a new demographic dimension. More important, unlike previous studies, our work focuses on investigating how different demographic groups (i.e. male/female, asian/black/white) connect with other.</p><h3>Demographics and Linguistic Analysis</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*mFgZVHReAmO6gfUE." /><figcaption>“An open book in Ukrainian seen through a pair of glasses” by <a href="https://unsplash.com/@ratushny?utm_source=medium&amp;utm_medium=referral">Dmitry Ratushny</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>In the field of demographics, most studies use linguistic analysis in order to extract useful features for predicting demographic information as gender, race, and age. Burger <em>et al.</em> [17] produce <em>ngrams</em> from users’ tweets, description, screen name, and full name, in order to predict Twitter user gender. They conclude that the training of an SVM classifier with the combination of all factors can create an efficient and accurate prediction scheme (92% accuracy) for gender classification. Also, Chen <em>et al.</em> [23] introduce a similar methodology for predicting gender, ethnicity, and age. However, using ngrams from the social neighbors, including followers and friends, and the distribution of 100 generated topics of LDA algorithm as the input of SVM classifier. In their results, the performance of classification is much lower in terms of ethnicity and age. Gilbert <em>et al.</em> [37] present an interesting statistical overview in Twitter and Pinterest using textual analysis and comparing what users write on Pinterest to what write text in Twitter. Cunha <em>et al.</em> [38] used Twitter data to analyze the difference between males and females in terms of generation of hashtags. Their results emphasize gender as a factor that influences the user’s choice of specific hashtags to a specific topic.</p><h3><strong>Algorithmic and Data Transparency</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*TZn6X3a1K0NYEkir." /><figcaption>Photo by <a href="https://unsplash.com/@igorovsyannykov?utm_source=medium&amp;utm_medium=referral">Igor Ovsyannykov</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Demographics is also an important aspect of providing transparency. Therefore, demographic distribution of users may be leveraged to make other crowdsourced systems (e.g., social search [15] and Twitter trending topic analysis [27]) transparent as well. For algorithms that operate on large-scale crowdsourced data, it is paramount to make the outputs of the algorithms (and the algorithm itself) transparent, and understand the non-uniformities in the inputs to the algorithms.</p><p>Increasingly, researchers and governments are recognizing the importance of making algorithms transparent. The White House recently released a report that concludes that the practitioners must ensure that AI-enabled systems are open, transparent, and understandable [39]. In news production, black-box automated writing bots or any other difficult-to-parse algorithmic systems used to produce or curate news may represent infringements on the basic ethics of journalism [40].</p><p>Indeed, the controversy around Facebook using human editors on their trending topics teaches a lesson about the importance of transparency. On one hand, when humans were editing trending topics, they were accused to select and filter content [41]. On the other hand, when humans were removed from the process, Facebook was accused of featuring fake news as trending [42]. Recently, Twitter announced the use of deep learning in their timeline algorithm and how it works in order to make it transparent to the public [43].</p><h3>Recommendation Diversity</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*OE2JlZkL33PJ4ksl." /><figcaption>“Artist dips paintbrush in bucket of paint among other pots of different colored paint” by <a href="https://unsplash.com/@russn_fckr?utm_source=medium&amp;utm_medium=referral">russn_fckr</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Diversity is increasingly recognized as an important metric for evaluating the effectiveness of online recommendations [44]. There have been multiple attempts in the past to introduce diversity in content search and recommendation systems. Gollapudi <em>et al.</em> [45] and Agrawal <em>et al.</em> [46] explored several approaches for search result diversification. Ziegler <em>et al.</em> [47] proposed algorithms to introduce <em>Novelty and Serendipity</em> in collaborative filtering based recommendations. Basically, search result diversification suggests a trade-off between showing diverse results in the top position for a given query and have more appropriate results that approximate to the user intent [48, 49]. There is also an effort to study stereotypes in Google and Bing search engines. Araújo <em>et al.</em> [50] query the search engines for beauty and ugly females. Their results show the existence of both negative stereotype for black females and positive stereotype for white females in terms of beauty. They also noticed that there are negative stereotypes about older females in U.S. while in hispanic countries share negative stereotypes about black females, positive for whites and neutral for asians. Sheth <em>et al.</em> [51] propose the “Social Diversity”, a new approach to diversity in recommendations that uses social network information to diversify recommendation results.</p><p>In 2015, Google search engine suffered from a lack of diversity when their search algorithm labeled black people as “gorillas” [52, 53]. It reinforces the importancy of using demographic information to provide diversity in recommendation systems. More recently, in 2016, a Microsoft Twitter bot that uses artificial intelligence quickly became a racist [54]. It also highlights the need for a better understanding of designing systems. Depending on which kind of output the algorithm gives to someone it could ruin her life.</p><h3>Fairness</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*3IVANGsyIu2PSRCY." /><figcaption>Photo by <a href="https://unsplash.com/@heftiba?utm_source=medium&amp;utm_medium=referral">Toa Heftiba</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Some countries have anti-discrimination laws in order to prohibit unfair treatment of individuals based on sensitive attributes (e.g. race and gender). It can be hard to identify the source of the problem or to explain it to a court [55]. Moreover, in 2013 the New York Police Department (<a href="http://www1.nyc.gov/site/nypd/index.page">NYPD</a>) started the Stop-question-and-frisk program (<a href="https://en.wikipedia.org/wiki/Stop-and-frisk_in_New_York_City">SQF</a>) to reduce criminality. It consists of temporally detaining, questioning, and at times searching civilians on the street for weapons and other contraband. Goel <em>et al.</em> [56] found that blacks and hispanics make up more than 80% of individuals stopped by SQF, even though they constitute approximately 50% of the New York City population.</p><p>Likewise, labor discrimination is also a difficult problem that has been studied for many years [57]. There are some studies that uncovered discrimination in traditional labour markets [58 , 59, 60]. More recently, Hannák <em>et al.</em> [19] explore the correlations between two freelance marketplaces, <a href="https://www.taskrabbit.com/">TaskRabbit</a> and <a href="https://www.fiverr.com/">Fiverr</a>, and quantify race- and gender-biases. They found that on TaskRabbit females, especially white females, receive 10% fewer than males with equivalent work experience. Black workers, especially males, receive significantly lower feedback scores than others. On the other hand, on Fiverr, black workers, especially males, receive ~32% fewer reviews than other males. Asian workers, also especially males, receive higher scores than others. Finally, their linguistic analysis shows that reviews for black females include fewer positive adjectives, while for blacks (in general) more negatives. Another, study shows that black tenants have much fewer chances of renting a place on <a href="https://www.airbnb.com/">Airbnb</a> [57]. Also, some drivers for both <a href="https://www.uber.com/">UberX</a> and <a href="https://www.lyft.com/">Lyft</a> discriminate on the basis of the perceived race of the traveler [61]. Google served ads that despised African-Americans [62], whilst Google did not show ads for high-paying jobs from females [63]. Further, Black and female sellers earn less than white and male sellers on <a href="https://www.ebay.com/">eBay</a> [64 , 65].</p><p>A number of studies found racial disparities in automated [66] as well as human [56, 67] decision making systems related to criminal justice. Recent studies focus on designing decision making systems that avoid unfairness [68, 69, 70, 71], proposing methods for detecting [68, 69, 72, 73] and removing [74, 68, 75, 76, 77, 72, 70, 71] unfairness. Furthermore, Zafar <em>et al.</em> [78] propose fair classifier formulations in order to remove disparate only on false-positive and false-negative rates. They provide a flexible trade-off between disparate mistreatment-based fairness and accuracy.</p><h3>Conclusion</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Js0S6RNNQJVVCMxV." /><figcaption>Photo by <a href="https://unsplash.com/@johnschno?utm_source=medium&amp;utm_medium=referral">John Schnobrich</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Exploring demographic aspects is important both from the perspective of systems and from the sociological point of view. Demographics is thekey to providing more fairness algorithm and classifiers. I hope this article may help developers and the research community to improve their systems in order to provide tools to decrease racial and gender disparities.</p><p>I developed and deployed the <a href="http://twitter-app.mpi-sws.org/who-makes-trends"><em>Who Makes Trends?</em></a> Web-based service in order to make the demographic biases in Twitter trends more transparent and show how powerful are demographic aspects in providing transparency. I believe this strategy can also be applied in different systems such as Google Suggestions and it would help us to study stereotype, reappropriation, and many other social concepts.</p><p>Please, feel free to comment and send this article to your friends and researchers.</p><p>For a more detailed information and my published work: <a href="http://johnnatan.me/">http://johnnatan.me</a></p><p>Thank you so much :)</p><h3>About me</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/1*vKxMKObPxWG27V_mQXqSVg.jpeg" /><figcaption>Johnnatan Messias — Master in Computer Science</figcaption></figure><p>Currently, I am a Data Scientist and Full Stack Software Engineer at <a href="http://kunumi.ai/">Kunumi</a>. As a Data Scientist, I develop machine learning systems to predict behaviors and states, such as forecasting delinquency, risk, and other prediction applications. On the another hand, as a Full Stack Software Engineer, I perform tasks ranging from designing architectures to deploying and maintaining moving parts of the company’s core software.</p><p>I received a Master degree in Computer Science from Universidade Federal de Minas Gerais (<a href="http://www.ufmg.br/">UFMG</a>) with Dr. Fabricio Benevenuto as my advisor. During my MsC, I also held research intern position at Max Planck Institute for Software Systems (<a href="http://www.mpi-sws.org/">MPI-SWS</a>) in Saarbrücken/Germany. I studied Computer Science at Eötvös Loránd University (<a href="https://www.elte.hu/en/">ELTE</a>) in Budapest/Hungary in 2013/2014 during my exchange program (Science Without Borders — CAPES). I received a bachelor degree in Computer Science from Universidade Federal de Ouro Preto (<a href="http://www.ufop.br/">UFOP</a>), in 2015.</p><h3><strong>References</strong></h3><p>[1] — <a href="https://www.statista.com/statistics/264810/number-of-monthly-active-facebook-users-worldwide/">https://www.statista.com/statistics/264810/number-of-monthly-active-facebook-users-worldwide/</a></p><p>[2] — <a href="https://www.statista.com/statistics/282087/number-of-monthly-active-twitter-users/">https://www.statista.com/statistics/282087/number-of-monthly-active-twitter-users/</a></p><p>[3] — Dutta, S. (2010). What’s your personal social media strategy. Harvard business review, 88(11):127–130.</p><p>[4] — Ronson, J. (2016). So you’ve been publicly shamed. Riverhead Books (Hardcover).</p><p>[5] — Blevins, C. and Mullen, L. (2015). Jane, John… Leslie? a historical method for algorithmic gender prediction. Digital Humanities Quarterly, 9(3).</p><p>[6] — Karimi, F., Wagner, C., Lemmerich, F., Jadidi, M., and Strohmaier, M. (2016). Inferring gender from names on the web: A comparative evaluation of gender detection methods. In Proceedings of the 25th International Conference on World Wide Web.</p><p>[7] — Liu, W. and Ruths, D. (2013). What’s in a name? Using first names as features for gender inference in Twitter. In AAAI Spring Symposium Series, volume 13, page 01.</p><p>[8] — Mislove, A., Lehmann, S., Ahn, Y.-Y., Onnela, J.-P., and Rosenquist, J. N. (2011). Understanding the demographics of Twitter users. In In Proceedings of the 5th International AAAI Conference on Weblogs and Social Media.</p><p>[9] — Nilizadeh, S., Groggel, A., Lista, P., Das, S., Ahn, Y.-Y., Kapadia, A., and Rojas, F. (2016). Twitter’s glass ceiling: The effect of perceived gender on online visibility. In In Proceedings of the 10th International AAAI Conference on Weblogs and Social Media.</p><p>[10] — Cotter, D. A., Hermsen, J. M., Ovadia, S., and Vanneman, R. (2001). The glass ceiling effect. Social forces, 80(2):655–681.</p><p>[11] — Bollen, J., Mao, H., and Zeng, X. (2011). Twitter mood predicts the stock market. Journal of Computational Science, 2(1):1–8.</p><p>[12] — Messias, J., Benevenuto, F., Weber, I., and Zagheni, E. (2016). From migration corridors to clusters: The value of Google+ data for migration studies. In Proceedings of the 2016 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining.</p><p>[13] — O’Connor, B., Balasubramanyan, R., Routledge, B. R., and Smith, N. A. (2010). From tweets to polls: Linking text sentiment to public opinion time series. In Proceedings of the 4th International AAAI Conference on Weblogs and Social Media.</p><p>[14] — Tumasjan, A., Sprenger, T. O., Sandner, P. G., and Welpe, I. M. (2010). Predicting elections with twitter: What 140 characters reveal about political sentiment. In Proceedings of the 4th International AAAI Conference on Web and Social Media.</p><p>[15] — Kulshrestha, J., Eslami, M., Messias, J., Zafar, M. B., Ghosh, S., Gummadi, K. P., and Karahalios, K. (2017). Quantifying search bias: Investigating sources of bias for political searches in social media. In Proceedings of the 20th ACM Conference on Computer-Supported Cooperative Work and Social Computing.</p><p>[16] — Pennacchiotti, M. and Popescu, A.-M. (2011). A machine learning approach to Twitter user classification. In Proceedings of the 5th International AAAI Conference on Web and Social Media.</p><p>[17] — Burger, J. D., Henderson, J., Kim, G., and Zarrella, G. (2011). Discriminating gender on Twitter. In Proceedings of the Conference on Empirical Methods in Natural Language Processing, pages 1301–1309. Association for Computational Linguistics.</p><p>[18] — Wachs, J., Hannak, A., Voros, A., and Daroczy, B. (2017). Why do men get more attention? Exploring factors behind success in an online design community. In Proceedings of the 11th International AAAI Conference on Web and Social Media, ICWSM’17, Montreal, Canada.</p><p>[19] — Hannák, A., Wagner, C., Garcia, D., Mislove, A., Strohmaier, M., and Wilson, C. (2017). Bias in online freelance marketplaces: Evidence from Taskrabbit and Fiverr. In Proceedings of the 2017 ACM Conference on Computer Supported Cooperative Work and Social Computing, CSCW’17, pages 1914–1933, New York, NY, USA. ACM.</p><p>[20] — Messias, J., Vikatos, P., and Benevenuto, F. (2017). White, man, and highly followed: Gender and race inequalities in Twitter. In Proceedings of the IEEE/WIC/ACM International Conference on Web Intelligence, WI’17, New York, NY, USA. ACM.</p><p>[21] — Vikatos, P., Messias, J., Miranda, M., and Benevenuto, F. (2017). Linguistic diversities of demographic groups in Twitter. In Proceedings of the 28th ACM Conference on Hypertext and Social Media. ACM.</p><p>[22] — Messias, J. Characterizing Interconnections and Linguistic Patterns in Twitter. Master Thesis in Computer Science. Universidade Federal de Minas Gerais — UFMG, 2017.</p><p>[23] — Chen, X., Wang, Y., Agichtein, E., and Wang, F. (2015). A comparative study of demographic attribute inference in Twitter. In Proceedings of the 9th International AAAI Conference on Weblogs and Social Media.</p><p>[24] — Culotta, A., Kumar, N. R., and Cutler, J. (2015). Predicting the demographics of twitter users from website traffic data. In Proceedings of the 29th AAAI Conference on Artificial Intelligence.</p><p>[25] — Reis, J. C. S., Kwak, H., An, J., Messias, J., and Benevenuto, F. (2017). Demographics of news sharing in the U.S. Twittersphere. In Proceedings of the 28th ACM Conference on Hypertext and Social Media. ACM.</p><p>[26] — An, J. and Weber, I. (2016). #greysanatomy vs. #yankees: Demographics and hashtag use on Twitter. In Proceedings of the 10th International AAAI Conference on Web and Social Media.</p><p>[27] — Chakraborty, A., Messias, J., Benevenuto, F., Ghosh, S., Ganguly, N., and Gummadi, K. P. (2017). Who makes trends? U demographic biases in crowdsourced recommendations. In Proceedings of the 11th International AAAI Conference on Web and Social Media.</p><p>[28] — McPherson, M., Smith-Lovin, L., and Cook, J. M. (2001). Birds of a feather: Homophily in social networks. Annual review of sociology, pages 415–444.</p><p>[29] — Willis, J. and Todorov, A. (2006). First impressions making up your mind after a 100-ms exposure to a face. Psychological science, 17(7):592–598.</p><p>[30] — Morales, A., Borondo, J., Losada, J. C., and Benito, R. M. (2014). Efficiency of human activity on information spreading on Twitter. Social Networks, 39:1–11.</p><p>[31] — Comarela, G., Crovella, M., Almeida, V., and Benevenuto, F. (2012). Understanding factors that affect response rates in Twitter. In Proceedings of the 23rd ACM Conference on Hypertext and Social Media.</p><p>[32] — Romero, D. M., Galuba, W., Asur, S., and Huberman, B. A. (2011). Influence and passivity in social media. In Joint European Conference on Machine Learning and Knowledge Discovery in Databases, pages 18–33. Springer.</p><p>[33] — Wu, S., Hofman, J. M., Mason, W. A., and Watts, D. J. (2011). Who says what to whom on Twitter. In Proceedings of the 20th International Conference on World Wide Web.</p><p>[34] — Freitas, C., Benevenuto, F., Ghosh, S., and Veloso, A. (2015). Reverse engineering socialbot infiltration strategies in Twitter. In Proceedings of the 2015 IEEEACM International Conference on Advances in Social Networks Analysis and Mining.</p><p>[35] — Graells-Garrido, E., Lalmas, M., and Menczer, F. (2015). First women, second sex: gender bias in Wikipedia. In Proceedings of the 26th ACM Conference on Hypertext &amp; Social Media.</p><p>[36] — Wagner, C., Graells-Garrido, E., Garcia, D., and Menczer, F. (2016). Women through the glass ceiling: gender asymmetries in Wikipedia. EPJ Data Science, 5(1):1.</p><p>[37] — Gilbert, E., Bakhshi, S., Chang, S., and Terveen, L. (2013). “I need to try this”?: A statistical overview of pinterest. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. ACM.</p><p>[38] — Cunha, E., Magno, G., Almeida, V., Gonçalves, M. A., and Benevenuto, F. (2012). A gender based study of tagging behavior in Twitter. In Proceedings of the 23rd ACM Conference on Hypertext and Social Media, HT ’12, pages 323–324, New York, NY, USA. ACM.</p><p>[39] — Felton, E. and Lyons, T. (2016). The administration’s report on the future of artificial intelligence. <a href="https://whitehouse.gov/blog/2016/10/12/administrations-report-future-artificial-intelligence.">https://whitehouse.gov/blog/2016/10/12/administrations-report-future-artificial-intelligence.</a></p><p>[40] — Diakopoulos, N. and Koliska, M. (2016). Algorithmic transparency in the news media. Digital Journalism.</p><p>[41] — Nunez, M. (2016). Former Facebook Workers: We Routinely Suppressed Conservative News. <a href="https://gizmodo.com/former-facebook-workers-we-routinely-suppressed-conser-1775461006.">https://gizmodo.com/former-facebook-workers-we-routinely-suppressed-conser-1775461006.</a></p><p>[42] — Ohlheiser, A. (2016). Three days after removing human editors, facebook is already trending fake news. washingtonpost.com/news/ the-intersect/wp/2016/08/29/a- fake-headline-about-megyn-kelly-was-trending-on-facebook/.</p><p>[43] — Koumchatzky, N. and Andryeyev, A. (2017). Using deep learning at scale in Twitter’s timelines. <a href="https://blog.twitter.com/engineering/en_us/topics/insights/2017/using-deep-learning-at-scale-in-twitters-timelines.html">https://blog.twitter.com/engineering/en_us/topics/insights/2017/using-deep-learning-at-scale-in-twitters-timelines.html</a><a href="http://blog.twitter.com/engineering/en_us/topics/insights/2017/using-deep-learning-at-scale-in-witters-timelines.html.">.</a></p><p>[44] — Chen, L., Wu, W., and He, L. (2016). Personality and Recommendation Diversity, pages 201–225. Springer International Publishing, Cham.</p><p>[45] — Gollapudi, S. and Sharma, A. (2009). An axiomatic approach for result diversification. In ACM WWW.</p><p>[46] — Agrawal, R., Gollapudi, S., Halverson, A., and Ieong, S. (2009). Diversifying search results. In ACM WSDM.</p><p>[47] — Ziegler, C.-N., McNee, S. M., Konstan, J. A., and Lausen, G. (2005). Improving recommendation lists through topic diversification. In ACM WWW.</p><p>[48] — Carbonell, J. and Goldstein, J. (1998). The use of mmr, diversity-based reranking for reordering documents and producing summaries. In Proceedings of the 21st annual international ACM SIGIR conference on Research and development in information retrieval, pages 335–336. ACM.</p><p>[49] — Chen, H. and Karger, D. R. (2006). Less is more: probabilistic models for retrieving fewer relevant documents. In Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval, pages 429–436. ACM.</p><p>[50] — Araújo, C. S., Meira Jr, W., and Almeida, V. (2016). Identifying stereotypes in the online perception of physical attractiveness. In International Conference on Social Informatics, pages 419–437. Springer.</p><p>[51] — Sheth, S. K., Bell, J. S., Arora, N., and Kaiser, G. E. (2011). Towards diversity in recommendations using social networks.</p><p>[52] — Guynn, J. (2015). Google photos labeled black people as “gorillas”. USA Today, July.</p><p>[53] — Barr, A. (2015). Google mistakenly tags black people as “gorillas”, showing limits of algorithms. The Wall Street Journal.</p><p>[54] — Victor, D. (2016). Microsoft created a Twitter bot to learn from users. It quickly became a racist jerk. The New York Times.</p><p>[55] — Barocas, S. and Selbst, A. D. (2016). Big data’s disparate impact. California Law Review, 104:671.</p><p>[56] — Goel, S., Rao, J. M., Shroff, R., et al. (2016). Precinct or prejudice? understanding racial disparities in New York City’s stop-and-frisk policy. The Annals of Applied Statistics, 10(1):365–394.</p><p>[57] — Todisco, M. (2014). Share and share alike: Considering racial discrimination in the nascent room-sharing economy. Stan. L. Rev. Online, 67:121–129.</p><p>[58] — Bertrand, M. and Mullainathan, S. (2004). Are Emily and Greg more employable than Lakisha and Jamal? A field experiment on labor market discrimination. The American Economic Review, 94(4):991–1013.</p><p>[59] — Carlsson, M. and Rooth, D.-O. (2007). Evidence of ethnic discrimination in the Swedish labor market using experimental data. Labour Economics, 14(4):716–729.</p><p>[60] — Dovidio, J. F. and Gaertner, S. L. (2000). Aversive racism and selection decisions: 1989 and 1999. Psychological science, 11(4):315–319.</p><p>[61] — Ge, Y., Knittel, C. R., MacKenzie, D., and Zoepf, S. (2016). Racial and gender discrimination in transportation network companies. Technical report, National Bureau of Economic Research.</p><p>[62] — Sweeney, L. (2013). Discrimination in online Ad delivery. Queue, 11(3):10:10–10:29. ISSN 1542–7730.</p><p>[63] — Datta, A., Tschantz, M. C., and Datta, A. (2015). Automated experiments on ad privacy settings. Proceedings on Privacy Enhancing Technologies, 2015(1):92–112.</p><p>[64] — Ayres, I., Banaji, M., and Jolls, C. (2015). Race effects on ebay. The RAND Journal of Economics, 46(4):891–917.</p><p>[65] — Kricheli-Katz, T. and Regev, T. (2016). How many cents on the dollar? Women and men in product markets. Science advances, 2(2):e1500599.</p><p>[66] — Angwin, J., Larson, J., Mattu, S., and Kirchner, L. (2016). Machine bias: There’s software used across the country to predict future criminals. And it’s biased against blacks. ProPublica, May, 23.</p><p>[67] — Ridgeway, G. and MacDonald, J. M. (2009). Doubly robust internal benchmarking and false discovery rates for detecting racial bias in police stops. Journal of the American Statistical Association, 104(486):661–668.</p><p>[68] — Feldman, M., Friedler, S. A., Moeller, J., Scheidegger, C., and Venkatasubramanian, S. (2015). Certifying and removing disparate impact. In Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD’15, pages 259–268, New York, NY, USA. ACM.</p><p>[69] — Luong, B. T., Ruggieri, S., and Turini, F. (2011). K-NN as an implementation of situation testing for discrimination discovery and prevention. In Proceedings of the 17th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD’11, pages 502–510, New York, NY, USA. ACM.</p><p>[70] — Zafar, M. B., Valera, I., Rodriguez, M. G., and Gummadi, K. P. (2017b). Fairness constraints: Mechanisms for fair classification. In Proceedings of the 20th International Conference on Artificial Intelligence and Statistics, AISTATS’17.</p><p>[71] — Zemel, R., Wu, Y., Swersky, K., Pitassi, T., and Dwork, C. (2013). Learning fair representations. In Proceedings of the 30th International Conference on Machine Learning, ICML’13, pages 325–333.</p><p>[72] — Pedreshi, D., Ruggieri, S., and Turini, F. (2008). Discrimination-aware data mining. In Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining, pages 560–568. ACM.</p><p>[73] — Romei, A. and Ruggieri, S. (2014). A multidisciplinary survey on discrimination analysis. The Knowledge Engineering Review, 29(05):582–638.</p><p>[74] — Dwork, C., Hardt, M., Pitassi, T., Reingold, O., and Zemel, R. (2012). Fairness through awareness. In Proceedings of the 3rd Innovations in Theoretical Computer Science Conference, pages 214–226. ACM.</p><p>[75] — Goh, G., Cotter, A., Gupta, M., and Friedlander, M. P. (2016). Satisfying real-world goals with dataset constraints. In Advances in Neural Information Processing Systems, pages 2415–2423.</p><p>[76] — Kamiran, F. and Calders, T. (2012). Data preprocessing techniques for classification without discrimination. Knowledge and Information Systems, 33(1):1–33.</p><p>[77] — Kamishima, T., Akaho, S., Asoh, H., and Sakuma, J. (2012). Fairness-aware classifier with prejudice remover regularizer. Machine Learning and Knowledge Discovery in Databases, pages 35–50.</p><p>[78] — Zafar, M. B., Valera, I., Gomez Rodriguez, M., and Gummadi, K. P. (2017a). Fairness beyond disparate treatment &amp; disparate impact: Learning classification without disparate mistreatment. In Proceedings of the 26th International Conference on World Wide Web, WWW’17, pages 1171–1180, Republic and Canton of Geneva, Switzerland. International World Wide Web Conferences Steering Committee.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ca52790dfecf" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>