No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
De l’information aux industries culturelles, l’hypothèse chahutée de la bulle de filtre
This pedagogical toolkit complements the competency framework entitled “AI Ethics Training in Higher Education: Competency Framework", which we produced thanks to the "Pôle montréalais d'enseignement supérieur en intelligence artificielle" (PIA). In this pedagogical toolkit, we provide concrete ways of carrying out training activities that will help develop the various aspects of AI ethics competency. For each activity, you will find a short explanation of the link between the proposed activity and the elements of the competency framework. This pedagogical toolkit should be seen as a working document that can be used creatively by the teachers who use it.
Cette trousse pédagogique est complémentaire au référentiel de compétence intitulé "Former à l’éthique de l’IA en enseignement supérieur : référentiel de compétence" que nous avons produit grâce au Pôle montréalais d’enseignement supérieur en intelligence artificielle (PIA). Nous proposons dans cette trousse pédagogique des moyens concrets pour réaliser des activités de formation qui permettront de développer les différents aspects de la compétence en éthique de l’IA. Pour chaque activité, vous retrouverez une courte explication du lien entre l’activité proposée et ces éléments provenant du référentiel de compétence. Cette trousse pédagogique doit être envisagée comme un document de travail à partir duquel la créativité des personnes enseignantes qui l’utiliseront pourra être mise à profit.
Computing Taste is a remarkable book about people who design and build commercial music recommendationsystems.
This article identifies five key themes, or sets of criticisms, that have emerged in online commentary on the new musical system centred on streaming platforms, and in related academic research: Streaming encourages ‘functional’ rather than meaningful, aesthetic musical experience. Streaming encourages bland, unchallenging music. Streaming makes musical experience passive and distracted, and music recedes into the background (here the article also discusses limitations of the widely used concepts of ‘ubiquitous music’ and ‘ubiquitous listening’). Streaming makes music tracks and songs shorter, and musical experience more fragmented. Streaming discourages and/or limits musical discovery and adventurousness.
The article addresses each of these themes in turn, examining the degree to which criticisms of streaming’s effects on musical experience along these lines might be considered valid, and the degree to which they might genuinely enhance critical understanding of contemporary musical experience. It also considers these themes in relation to older forms of evaluation, particularly those that developed in the 20th century in response to the industrialisation of music, and argues that many recent criticisms problematically reproduce older anxieties and assumptions.
The new high-choice media environment has raised concerns that users of social networking sites primarily select political information that supports their political opinions and avoid information that challenges them. This behaviour is reinforced by personalisation algorithms that create filter bubbles and both narrow the available content and exclude challenging information over time. These concerns have, however, been contested. This article challenges the underlying theoretical assumptions about filter bubbles, and compares filter bubbles to what we already know about selective exposure and human psychology. The article lists nine counterarguments to the filter bubble thesis. In short, I argue that the assumptions of filter bubbles contradict many of the previous findings of selective exposure research. More specifically, when discussing filter bubbles there is a risk of confusing two arguments: one strong - but also trivial - that is about technology (e.g., personalisation leads to different information), and one weak and speculative - but also the most interesting - that is about society (e.g., personalisation increases political polarisation in society). © 2021 Nordicom and respective authors. This is an Open Access work licensed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International Public licence (CC BY-NC-ND 4.0). To view a copy of the licence, visit https://creativecommons.org/licenses/by-nc-nd/4.0/
Although human existence is enveloped by ideologies, remarkably little is understood about the relationships between ideological attitudes and psychological traits. Even less is known about how cognitive dispositions—individual differences in how information is perceived and processed— sculpt individuals' ideological worldviews, proclivities for extremist beliefs and resistance (or receptivity) to evidence. Using an unprecedented number of cognitive tasks ( n = 37) and personality surveys ( n = 22), along with data-driven analyses including drift-diffusion and Bayesian modelling, we uncovered the specific psychological signatures of political, nationalistic, religious and dogmatic beliefs. Cognitive and personality assessments consistently outperformed demographic predictors in accounting for individual differences in ideological preferences by 4 to 15-fold. Furthermore, data-driven analyses revealed that individuals’ ideological attitudes mirrored their cognitive decision-making strategies. Conservatism and nationalism were related to greater caution in perceptual decision-making tasks and to reduced strategic information processing, while dogmatism was associated with slower evidence accumulation and impulsive tendencies. Religiosity was implicated in heightened agreeableness and risk perception. Extreme pro-group attitudes, including violence endorsement against outgroups, were linked to poorer working memory, slower perceptual strategies, and tendencies towards impulsivity and sensation-seeking—reflecting overlaps with the psychological profiles of conservatism and dogmatism. Cognitive and personality signatures were also generated for ideologies such as authoritarianism, system justification, social dominance orientation, patriotism and receptivity to evidence or alternative viewpoints; elucidating their underpinnings and highlighting avenues for future research. Together these findings suggest that ideological worldviews may be reflective of low-level perceptual and cognitive functions.
This article is part of the theme issue ‘The political brain: neurocognitive and computational mechanisms’.
Diversity is a concept relevant to numerous domains of research varying from ecology, to information theory, and to economics, to cite a few. It is a notion that is steadily gaining attention in the information retrieval, network analysis, and artificial neural networks communities. While the use of diversity measures in network-structured data counts a growing number of applications, no clear and comprehensive description is available for the different ways in which diversities can be measured. In this article, we develop a formal framework for the application of a large family of diversity measures to heterogeneous information networks (HINs), a flexible, widely-used network data formalism. This extends the application of diversity measures, from systems of classifications and apportionments, to more complex relations that can be better modeled by networks. In doing so, we not only provide an effective organization of multiple practices from different domains, but also unearth new observables in systems modeled by heterogeneous information networks. We illustrate the pertinence of our approach by developing different applications related to various domains concerned by both diversity and networks. In particular, we illustrate the usefulness of these new proposed observables in the domains of recommender systems and social media studies, among other fields.
The role of recommendation algorithms in online user confinement is at the heart of a fast-growing literature. Recent empirical studies generally suggest that filter bubbles may principally be observed in the case of explicit recommendation (based on user-declared preferences) rather than implicit recommendation (based on user activity). We focus on YouTube which has become a major online content provider but where confinement has until now been little-studied in a systematic manner. We aim to contribute to the above literature by showing whether recommendation on YouTube exhibits phenomena typical of filter bubbles, tending to lower the diversity of consumed content. Starting from a diverse number of seed videos, we first describe the properties of the sets of suggested videos in order to design a sound exploration protocol able to capture latent recommendation graphs recursively induced by these suggestions. These graphs form the background of potential user navigations along non-personalized recommendations. From there, be it in topological, topical or temporal terms, we show that the landscape of what we call mean-field YouTube recommendations is often prone to confinement dynamics. Moreover, the most confined recommendation graphs i.e., potential bubbles, seem to be organized around sets of videos that garner the highest audience and thus plausibly viewing time.
L’article présente une analyse des réseaux d’échanges d’informations mis en place au sein de Facebook par les principales communautés politiques françaises lors de la campagne présidentielle de 2017. Il montre que ces communautés développent des logiques de partage d’information profondément homophiles, sur un plan idéologique, dans la mesure où elles font majoritairement circuler au sein de leurs pages Facebook des articles produits par des sources d’information alignées sur leurs orientations politiques respectives. Seuls les groupes liés à l’extrême droite se démarquent en suivant une logique davantage susceptible de polariser les opinions et de renforcer les convictions. Ils développent en effet des pratiques d’information qui agencent une bulle de désinformation où circulent en boucle de nombreux contenus produits par des sources d’information peu fiables. Nous interprétons ces résultats en défendant l’idée selon laquelle les groupes politiques tendent à s’inscrire dans ce type de processus de polarisation et de désinformation quand ils s’approprient Facebook pour échanger des points de vue critiques et contre-hégémoniques en espérant ainsi atteindre le centre du débat public en contournant les médias mainstream.
Due to their success, social network platforms are considered today as a major communication mean. In order to increase user engagement, they rely on recommender systems to personalize individual experience by filtering messages according to user interest and/or neighborhood. However some recent results exhibit that this personalization of content might increase the echo chamber effect and create filter bubbles. These filter bubbles restrain the diversity of opinions regarding the recommended content. In this paper, we first realize a thorough study of communities on a large Twitter dataset to quantify how recommender systems affect users’ behavior and create filter bubbles. Then we propose the Community Aware Model (CAM) to counter the impact of different recommender systems on information consumption. Our results show that filter bubbles concern up to 10% of users and our model based on similarities between communities enhance recommender systems.
Background
Digital spaces, and in particular social networking sites, are becoming increasingly present and influential in the functioning of our democracies. In this paper, we propose an integrated methodology for the data collection, the reconstruction, the analysis and the visualization of the development of a country’s political landscape from Twitter data.
Method
The proposed method relies solely on the interactions between Twitter accounts and is independent of the characteristics of the shared contents such as the language of the tweets. We validate our methodology on a case study on the 2017 French presidential election (60 million Twitter exchanges between more than 2.4 million users) via two independent methods: the comparison between our automated political categorization and a human categorization based on the evaluation of a sample of 5000 profiles descriptions; the correspondence between the reconfigurations detected in the reconstructed political landscape and key political events reported in the media. This latter validation demonstrated the ability of our approach to accurately reflect the reconfigurations at play in the off-line political scene.
Results
We built on this reconstruction to give insights into the opinion dynamics and the reconfigurations of political communities at play during a presidential election. First, we propose a quantitative description and analysis of the political engagement of members of political communities. Second, we analyze the impact of political communities on information diffusion and in particular on their role in the fake news phenomena. We measure a differential echo chamber effect on the different types of political news (fake news, debunks, standard news) caused by the community structure and emphasize the importance of addressing the meso-structures of political networks in understanding the fake news phenomena.
Conclusions
Giving access to an intermediate level, between sociological surveys in the field and large statistical studies (such as those conducted by national or international organizations) we demonstrate that social networks data make it possible to qualify and quantify the activity of political communities in a multi-polar political environment; as well as their temporal evolution and reconfiguration, their structure, their alliance strategies and their semantic particularities during a presidential campaign through the analysis of their digital traces. We conclude this paper with a comment on the political and ethical implications of the use of social networks data in politics. We stress the importance of developing social macroscopes that will enable citizens to better understand how they collectively make society and propose as example the “Politoscope”, a macroscope that delivers some of our results in an interactive way.
This article focuses on the various apparatus that guide the Internet user confronted with an ample broadcasting supply; it distinguishes between apparatus based on judgments and apparatus based on behaviour. It highlights the fact that the second type are becoming more and more important given their amazing efficiency due to precise analysis of the users’ behaviour. Each in its own manner, various forms of recommendation pose challenges to audio-visual online operators who will need to respond with new ideas for competitive differentiation. The personalised algorithmic recommendation, in particular, is transforming economic models, market structures and the organisation of professions.
In a high-choice media environment, there are fears that individuals will select media and content that reinforce their existing beliefs and lead to segregation based on interest and/or partisanship. This could lead to partisan echo chambers among those who are politically interested and could contribute to a growing gap in knowledge between those who are politically interested and those who are not. However, the high-choice environment also allows individuals, including those who are politically interested, to consume a wide variety of media, which could lead them to more diverse content and perspectives. This study examines the relationship between political interest as well as media diversity and being caught in an echo chamber (measured by five different variables). Using a nationally representative survey of adult internet users in the United Kingdom (N = 2000), we find that those who are interested in politics and those with diverse media diets tend to avoid echo chambers. This work challenges the impact of echo chambers and tempers fears of partisan segregation since only a small segment of the population are likely to find themselves in an echo chamber. We argue that single media studies and studies which use narrow definitions and measurements of being in an echo chamber are flawed because they do not test the theory in the realistic context of a multiple media environment.
Abstract: Recommender system is a growing proliferation in today online applications contributed to the problems of Information overloading. In a day to day life enormous amount of data is generated and collected leads to a problem of information overloading. This paper focuses on how to deal with the problem of information overloading and how to recommend an additional product to the end user using collaborative filtering (CF) recommendation algorithms. The personalized recommendation algorithm with their benefits and limitations are described. A pitfall occurs in CF recommendation system is described. An outline framework is proposed for the initial stage of recommendation. A sample problem statement is framed to show how recommendation technology is embedded in to the framework.
Keywords: Web Mining, Collaborative filtering, CF Algorithms, CF Framework, E-Commerce
Instead of purchasing individual content, streaming adopters rent access to libraries from which they can consume content at no additional cost. In this paper, we study how the adoption of music streaming affects listening behavior. Using a unique panel data set of individual consumers’ listening histories across many digital music platforms, adoption of streaming leads to very large increases in the quantity and diversity of consumption in the first months after adoption. Although the effects attenuate over time, even after half a year, adopters play substantially more, and more diverse, music. Relative to music ownership, where experimentation is expensive, adoption of streaming increases new music discovery. While repeat listening to new music decreases, users’ best discoveries have higher play rates. We discuss the implications for consumers and producers of music.
Data and the online appendix are available at https://doi.org/10.1287/mksc.2017.1051 .
Significance
We present evidence from five experiments in two countries suggesting the power and robustness of the search engine manipulation effect (SEME). Specifically, we show that ( i ) biased search rankings can shift the voting preferences of undecided voters by 20% or more, ( ii ) the shift can be much higher in some demographic groups, and ( iii ) such rankings can be masked so that people show no awareness of the manipulation. Knowing the proportion of undecided voters in a population who have Internet access, along with the proportion of those voters who can be influenced using SEME, allows one to calculate the win margin below which SEME might be able to determine an election outcome.
Exposure to news, opinion and civic information increasingly occurs through social media. How do these online networks influence exposure to perspectives that cut across ideological lines? Using de-identified data, we examined how 10.1 million U.S. Facebook users interact with socially shared news. We directly measured ideological homophily in friend networks, and examine the extent to which heterogeneous friends could potentially expose individuals to cross-cutting content. We then quantified the extent to which individuals encounter comparatively more or less diverse content while interacting via Facebook's algorithmically ranked News Feed, and further studied users' choices to click through to ideologically discordant content. Compared to algorithmic ranking, individuals' choices about what to consume had a stronger effect limiting exposure to cross-cutting content.
Copyright © 2015, American Association for the Advancement of Science.
Social media have quickly become a prevalent channel to access information,
spread ideas, and influence opinions. However, it has been suggested that
social and algorithmic filtering may cause exposure to less diverse points of
view, and even foster polarization and misinformation. Here we explore and
validate this hypothesis quantitatively for the first time, at the collective
and individual levels, by mining three massive datasets of web traffic, search
logs, and Twitter posts. Our analysis shows that collectively, people access
information from a significantly narrower spectrum of sources through social
media and email, compared to search. The significance of this finding for
individual exposure is revealed by investigating the relationship between the
diversity of information sources experienced by users at the collective and
individual level. There is a strong correlation between collective and
individual diversity, supporting the notion that when we use social media we
find ourselves inside "social bubbles". Our results could lead to a deeper
understanding of how technology biases our exposure to new information.
Faced with algorithmic governmentality
Algorithmic governmentality is characterized primarily by the following dual movement: a) abandoning all forms of “scale”, “benchmark”, or hierarchy, in favour of an immanent normativity evolving in real time, from which a “dual statistics” of the world emerges and which seems to do away with the old hierarchies devised by normal or average people; and b) avoiding all confrontation with individuals, whose opportunities for subjectification have become increasingly scarce.
This dual movement seems to be the fruit of contemporary statistics’ focus on relations. We seek to assess the extent to which these two aspects of the “algorithmic governmentality” thereby outlined, with its sole reliance on relations, could facilitate, first, processes of individuation through relations (Simondon) and, second, the emergence of new forms of life through the plane of immanence overtaking the plane of organization (Deleuze-Guattari). Through this comparison with the main contemporary philosophies of relations, it thus appears that thinking about the evolution and processes of individuation through relations necessarily pertains to the “disparate” – a heterogeneity of orders of magnitude, a multiplicity of regimes of existence – which algorithmic governmentality precisely incessantly suppresses by enclosing (digitized) reality on itself. Algorithmic governmentality tends rather to foreclose such emancipation perspectives by centring individuation processes on the subjective monad.
This paper investigates political homophily on Twitter. Using a combination of machine learning and social network analysis we classify users as Democrats or as Republicans based on the political content shared. We then investigate political homophily both in the network of reciprocated and nonreciprocated ties. We find that structures of political homophily differ strongly between Democrats and Republicans. In general, Democrats exhibit higher levels of political homophily. But Republicans who follow official Republican accounts exhibit higher levels of homophily than Democrats. In addition, levels of homophily are higher in the network of reciprocated followers than in the nonreciprocated network. We suggest that research on political homophily on the Internet should take the political culture and practices of users seriously.
An innovative investigation of the inner workings of Spotify that traces the transformation of audio files into streamed experience.
Spotify provides a streaming service that has been welcomed as disrupting the world of music. Yet such disruption always comes at a price. Spotify Teardown contests the tired claim that digital culture thrives on disruption. Borrowing the notion of “teardown” from reverse-engineering processes, in this book a team of five researchers have playfully disassembled Spotify's product and the way it is commonly understood.
Spotify has been hailed as the solution to illicit downloading, but it began as a partly illicit enterprise that grew out of the Swedish file-sharing community. Spotify was originally praised as an innovative digital platform but increasingly resembles a media company in need of regulation, raising questions about the ways in which such cultural content as songs, books, and films are now typically made available online.
Spotify Teardown combines interviews, participant observations, and other analyses of Spotify's “front end” with experimental, covert investigations of its “back end.” The authors engaged in a series of interventions, which include establishing a record label for research purposes, intercepting network traffic with packet sniffers, and web-scraping corporate materials. The authors' innovative digital methods earned them a stern letter from Spotify accusing them of violating its terms of use; the company later threatened their research funding. Thus, the book itself became an intervention into the ethics and legal frameworks of corporate behavior.
The rise of the "information society" offers not only considerable peril but also great promise. Beset from all sides by a never-ending barrage of media, how can we ensure that the most accurate information emerges and is heeded? In this book, Cass R. Sunstein develops a deeply optimistic understanding of the human potential to pool information, and to use that knowledge to improve our lives. In an age of information overload, it is easy to fall back on our own prejudices and insulate ourselves with comforting opinions that reaffirm our core beliefs. Crowds quickly become mobs. The justification for the Iraq war, the collapse of Enron, the explosion of the space shuttle Columbia--all of these resulted from decisions made by leaders and groups trapped in "information cocoons," shielded from information at odds with their preconceptions. How can leaders and ordinary people challenge insular decision making and gain access to the sum of human knowledge? Stunning new ways to share and aggregate information, many Internet-based, are helping companies, schools, governments, and individuals not only to acquire, but also to create, ever-growing bodies of accurate knowledge. Through a ceaseless flurry of self-correcting exchanges, wikis, covering everything from politics and business plans to sports and science fiction subcultures, amass--and refine--information. Open-source software enables large numbers of people to participate in technological development. Prediction markets aggregate information in a way that allows companies, ranging from computer manufacturers to Hollywood studios, to make better decisions about product launches and office openings. Sunstein shows how people can assimilate aggregated information without succumbing to the dangers of the herd mentality--and when and why the new aggregation techniques are so astoundingly accurate. In a world where opinion and anecdote increasingly compete on equal footing with hard evidence, the on-line effort of many minds coming together might well provide the best path to infotopia.
How companies like Amazon and Netflix know what “you might also like”: the history, technology, business, and social impact of online recommendation engines.
Increasingly, our technologies are giving us better, faster, smarter, and more personal advice than our own families and best friends. Amazon already knows what kind of books and household goods you like and is more than eager to recommend more; YouTube and TikTok always have another video lined up to show you; Netflix has crunched the numbers of your viewing habits to suggest whole genres that you would enjoy. In this volume in the MIT Press's Essential Knowledge series, innovation expert Michael Schrage explains the origins, technologies, business applications, and increasing societal impact of recommendation engines, the systems that allow companies worldwide to know what products, services, and experiences “you might also like.”
Schrage offers a history of recommendation that reaches back to antiquity's oracles and astrologers; recounts the academic origins and commercial evolution of recommendation engines; explains how these systems work, discussing key mathematical insights, including the impact of machine learning and deep learning algorithms; and highlights user experience design challenges. He offers brief but incisive case studies of the digital music service Spotify; ByteDance, the owner of TikTok; and the online personal stylist Stitch Fix. Finally, Schrage considers the future of technological recommenders: Will they leave us disappointed and dependent—or will they help us discover the world and ourselves in novel and serendipitous ways?
Ce livre propose une mise en perspective des grandes théories surveillancielles, qu'elles soient d'ordre politique, économique, philosophique ou médiatique,. Il vise à clarifier pour mieux les comprendre un certain nombre de notions apparentées, et parfois mêlées : surveillance, contrôle, sécurité, transparence, visibilité, exposition...
Il accorde enfin une grande importance aux textes "littéraires" de la surveillance qui accompagnent et précèdent parfois les grandes modèles théoriques.
The different technological affordances and revenue models of subscriber-funded, internet-distributed video streaming services have altered the competitive environments of audiovisual services. One category of these services, multinational SVODs (subscription video on demand), are changing the dynamics of transnational video distribution. Although having subscribers and offices, and commissioning content from many countries, are obvious measures of these services’ multinational status, the extent to which the distinct affordances of these services diminish the national lens through which all other international television trade occurs may be the most profound measure. The article explores how this too becomes a distinguishing competitive tool for Netflix that enables uncommon content strategies, such as the ability to program for tastes and sensibilities too small to effectively form a viable market for services limited by national reach.
Music streaming services provide people with access to vast libraries of music, but also encourage certain patterns of consumption. In this article, I use Spotify as a case and investigate the action potentials for exploring and archiving music. The personal role of music implies we may expect the ‘will to archive’ to be prevalent even if these archives are not based on individual ownership. First, an analysis of Spotify suggests that the machine agency of Spotify pushes people towards exploring music, whereas archiving features are material and depend on human action. Spotify is hence skewed towards prompting users to explore rather than archive music. Next, an analysis of 23 focus-group interviews suggests that users value opportunities to explore music, yet their practices are equally directed towards archiving music. Theoretically, this article delineates how objects with machine agency are different from material objects in terms of affordances. The action potentials of material objects are symmetrically constituted by what the objects provide relative to an active being. The action potentials of objects with machine agency interfere with this symmetry: the machine is designed to act on behalf of the human being, making certain affordances more perceivable than others.
Les services d’écoute musicale en streaming se situent à mi-chemin entre deux dispositifs antérieurs, la radio et la collection de disques, et deux modes d’agentivité associés : le consommateur pris en charge par la programmation musicale dans le cas de la radio, l’amateur dans le cas de la collection. Ils proposent ainsi un régime hybride de prise en charge personnalisée, adaptée aux goûts des usagers, mesurés par leurs écoutes passées, via l’usage de systèmes de recommandation. Dans quelle mesure la figure dessinée par les interfaces se traduit-elle dans les comportements réels des auditeurs ? Nous nous appuyons sur l’analyse des traces d’activités musicales d’un échantillon aléatoire de 4000 utilisateurs d’une plateforme musicale, suivi pendant une période de cinq mois, pour répondre à cette question. Nous décrivons d’abord la distribution des écoutes, et montrons que si la diversité des consommations sur la plateforme est plus grande que sur d’autres dispositifs, les écoutes demeurent fortement concentrées sur les artistes les plus populaires. Nous examinons ensuite les usages des recommandations par les usagers, en montrant qu’ils restent mesurés, inégalement distribués, et que les usages de découverte musicale, s’ils orientent les usagers vers des musiques de la mid-tail et de la long-tail, demeurent marginaux.
Finding facts about fake news
There was a proliferation of fake news during the 2016 election cycle. Grinberg et al. analyzed Twitter data by matching Twitter accounts to specific voters to determine who was exposed to fake news, who spread fake news, and how fake news interacted with factual news (see the Perspective by Ruths). Fake news accounted for nearly 6% of all news consumption, but it was heavily concentrated—only 1% of users were exposed to 80% of fake news, and 0.1% of users were responsible for sharing 80% of fake news. Interestingly, fake news was most concentrated among conservative voters.
Science , this issue p. 374 ; see also p. 348
A firsthand account and incisive analysis of modern protest, revealing internet-fueled social movements' greatest strengths and frequent challenges. To understand a thwarted Turkish coup, an anti-Wall Street encampment, and a packed Tahrir Square, we must first comprehend the power and the weaknesses of using new technologies to mobilize large numbers of people. An incisive observer, writer, and participant in today's social movements, Zeynep Tufekci explains in this accessible and compelling book the nuanced trajectories of modern protests-how they form, how they operate differently from past protests, and why they have difficulty persisting in their long-term quests for change. Tufekci speaks from direct experience, combining on-the-ground interviews with insightful analysis. She describes how the internet helped the Zapatista uprisings in Mexico, the necessity of remote Twitter users to organize medical supplies during Arab Spring, the refusal to use bullhorns in the Occupy Movement that started in New York, and the empowering effect of tear gas in Istanbul's Gezi Park. These details from life inside social movements complete a moving investigation of authority, technology, and culture-and offer essential insights into the future of governance.
Following the 2016 US presidential election, many have expressed concern about the effects of false stories ("fake news"), circulated largely through social media. We discuss the economics of fake news and present new data on its consumption prior to the election. Drawing on web browsing data, archives of fact-checking websites, and results from a new online survey, we find: 1) social media was an important but not dominant source of election news, with 14 percent of Americans calling social media their "most important" source; 2) of the known false news stories that appeared in the three months before the election, those favoring Trump were shared a total of 30 million times on Facebook, while those favoring Clinton were shared 8 million times; 3) the average American adult saw on the order of one or perhaps several fake news stories in the months around the election, with just over half of those who recalled seeing them believing them; and 4) people are much more likely to believe stories that favor their preferred candidate, especially if they have ideologically segregated social media networks.
Diversification has become one of the leading topics of recommender system research not only as a way to solve the over-fitting problem but also an approach to increasing the quality of the user's experience with the recommender system. This article aims to provide an overview of research done on this topic from one of the first mentions of diversity in 2001 until now. The articles ,and research, have been divided into three sub-topics for a better overview of the work done in the field of recommendation diversification: the definition and evaluation of diversity; the impact of diversification on the quality of recommendation results and the development of diversification algorithms themselves. In this way, the article aims both to offer a good overview to a researcher looking for the state-of-the-art on this topic and to help a new developer get familiar with the topic.
YouTube represents one of the largest scale and most sophisticated industrial recommendation systems in existence. In this paper, we describe the system at a high level and focus on the dramatic performance improvements brought by deep learning. The paper is split according to the classic two-stage information retrieval dichotomy: first, we detail a deep candidate generation model and then describe a separate deep ranking model. We also provide practical lessons and insights derived from designing, iterating and maintaining a massive recommendation system with enormous user-facing impact.
De nombreuses études ont démontré que la prise en compte du contexte améliore la qualité des systèmes de recommandation. Cependant, les méthodes traditionnelles permettent d'inférer le contexte à l'aide de données personnelles (localisation, date, âge, etc.). Dans ce papier, nous proposons de détecter automatiquement les changements de contexte, sans connaissance sur les utilisateurs (contexte explicite), mais en fonction des caractéristiques communes aux items consultés (contexte implicite). Pour ce faire, nous proposons un modèle formel capable d'établir une correspondance entre les variations de diversité au cours du temps dans les parcours des utilisateurs et les changements de contexte. Ce modèle a été testé sur un corpus musical de plus de 200.000 écoutes. Pour valider la pertinence de notre modèle, nous avons cherché à retrouver des événements à partir des changements de contexte détectés : notre modèle a ainsi permis de retrouver 88% des fins de sessions.
Online publishing, social networks, and web search have dramatically lowered the costs of producing, distributing, and discovering
news articles. Some scholars argue that such technological changes increase exposure to diverse perspectives, while others
worry that they increase ideological segregation. We address the issue by examining web-browsing histories for 50,000 US-located
users who regularly read online news. We find that social networks and search engines are associated with an increase in the
mean ideological distance between individuals. However, somewhat counterintuitively, these same channels also are associated
with an increase in an individual’s exposure to material from his or her less preferred side of the political spectrum. Finally,
the vast majority of online news consumption is accounted for by individuals simply visiting the home pages of their favorite,
typically mainstream, news outlets, tempering the consequences—both positive and negative—of recent technological changes.
We thus uncover evidence for both sides of the debate, while also finding that the magnitude of the effects is relatively
modest.
This article discusses the various algorithms that make up the Netflix recommender system, and describes its business purpose. We also describe the role of search and related algorithms, which for us turns into a recommendations problem as well. We explain the motivations behind and review the approach that we use to improve the recommendation algorithms, combining A/B testing focused on improving member retention and medium term engagement, as well as offline experimentation using historical member engagement data. We discuss some of the issues in designing and interpreting A/B tests. Finally, we describe some current areas of focused innovation, which include making our recommender system global and language aware.
This article discusses the first outcome results carried out in a study for the architecture of twelve French news site homepages. This particular analysis deals with identifying the different items that make up the different homepages, as well as, their quantification and the calculation of their relative area within the page. Each item has been taken into consideration as apart of a systematic and objective analysis in areas, concerning the type of content (journalistic, commercial...), the format (text, photo, video and audio), all in relation to it's actual positioning on the homepage. Here, the objective is to observe and understand the innovations by which the information has been treated and to highlight the uniqueness - or the differences between the media on the Internet and the traditional media itself. The results have revealed a high heritage of the written press (with predominance of written texts fixed pictures) as well as with the structures of homepages.
Eli Pariser coined the term 'filter bubble' to describe the potential for online personalization to effectively isolate people from a diversity of viewpoints or content. Online recommender systems - built on algorithms that attempt to predict which items users will most enjoy consuming - are one family of technologies that potentially suffers from this effect. Because recommender systems have become so prevalent, it is important to investigate their impact on users in these terms. This paper examines the longitudinal impacts of a collaborative filtering-based recommender system on users. To the best of our knowledge, it is the first paper to measure the filter bubble effect in terms of content diversity at the individual level. We contribute a novel metric to measure content diversity based on information encoded in user-generated tags, and we present a new set of methods to examine the temporal effect of recommender systems on the user experience. We do find that recommender systems expose users to a slightly narrowing set of items over time. However, we also see evidence that users who actually consume the items recommended to them experience lessened narrowing effects and rate items more positively.