Overview and details of the sessions of this conference. Please select a date or location to show only sessions at that day or location. Please select a single session for detailed view (with abstracts and downloads if available).
Trust in deconstructed recommender systems. Case study: News Recommender Systems
Marijn Martens1, Ralf De Wolf1, Bettina Berendt2, Lieven De Marez1
1imec-mict-UGent; 2KU Leuven
Increasingly, algorithms play an important role in everyday decision-making processes. Recommender systems, specifically, are algorithms that serve to influence end-users’ decision-making (e.g. what to read, who to befriend, who to rent to…). However, the companies that develop and produce these systems are not neutral, but have an economic goal and specific vision on how society should operate. These algorithms should thus never be trusted blindly.
An algorithm consists of collective human practices and consequently warm human and institutional choices. Therefore, they should be perceived as culture. Despite the many academics that are joining the debate to denounce the bias, opaqueness and unfairness often found in these algorithms, little empirical research has invested in treating algorithms in its socio-technical assembly as culture.
To better understand how end-users perceive these algorithmic systems, we strive to understand how they imagine and (dis)trust the different components of the socio-technical assembly. We are demystifying the imagined processes incorporated in these algorithmic systems in the minds of the end-user using a deconstructed version of Buchers’ (2017) algorithmic imaginary.
Currently, companies put ever more effort into personalizing news, using news recommender systems (NRS). NRS organize, select and aggregate news to influence the decision-making of an end-user without a transparent explanation on the process. Therefore, we focus our study on the end-users of these NRS.
In this qualitative study, we are interviewing 25 end-users of NRS to understand the assumptions and apprehend the (dis-)trust people have about the different elements of the socio-technical assembly of news recommender systems.
2:20pm - 2:40pm
API AND BEYOND: DETECTING COORDINATED BEHAVIOURS IN FACEBOOK INTERACTIONS AROUND POLITICAL NEWS STORIES
Fabio Giglietto, Nicola Righetti, Giada Marino
University of Urbino, Italy
This proposal is a follow-up of the project “Mapping Italian News Media Political Coverage in the Lead-up to 2018 General Election” (MINE). MINE aimed at creating a comprehensive map of the political news coverage created by the Italian online news media in the lead-up to 2018 general election.
The final report of the project highlighted how the populist narrative dominated the news (both in terms of volume of coverage and Facebook engagement), and pinpointed the diverging patterns of Facebook interactions employed by different partisan communities to amplify the reach of the contents aligned with their worldview by sharing the news stories on social media, while trying to reframe, through comments, the negative coverage of the party they support.
These insights led to further questions concerning the nature of the observed diverging patterns of Facebook interactions around political news. In particular, we wondered if the observed patterns were the result of a spontaneous grassroots effort or instead of a strategically organised attempt to manipulate the online news media landscape in order to game platforms algorithms in support of specific viewpoints, candidates and parties.
Data originally collected for MINE during 2018 via publically available Facebook API proved useful to identify the patterns, but fall short of providing compelling evidence on the nature of these behaviours. In order to shed some light on this question, we thus requested and obtained access to two additional datasets directly provided by Facebook and made available through the Social Science One (SSO) initiative.
2:40pm - 3:00pm
Trolling for engagement: Australian legacy news outlets seeking audience interaction metrics on Facebook through deliberately divisive content
Queensland University of Technology, Australia
This paper empirically investigates how two prominent Australian legacy news outlets – ABC News and News.com.au – operate according to what I term a social media logic of “engagement”, a concept which builds upon van Dijck & Poell’s notion of a social media logic of “popularity”. By a logic of engagement, I mean the necessity to maximize social media attention and interaction metrics. Rather than just valuing “popularity”, platforms instead place value on content that maximizes a multitude of feelings, sentiments, and reactions. Without sufficient engagement, outlets dependent on platforms such as Facebook are threatened by invisibility in the newsfeed. I specifically focus on the operations of ABC News and News.com.au on Facebook from 21 March 2018 – 10 April 2018. Within this period, I collected all the posts from each page, which amounted to 44 posts in total. From these posts, I strategically selected six posts of varying levels of engagement for closer qualitative analysis, with an emphasis on language and imagery. My findings in this paper suggest that the drive for monetizable and algorithmically-valued audience metrics on Facebook can encourage divisive and provocative news content that arouses strong negative feelings and promotes conflict. Trolls are those that deceive other users of their intentions, and seek to sow discord for their own purposes. Thus, it is beneficial to think about a potentially emerging practice of news “trolling”, as it appears that news outlets are adopting faux-naïve, and deliberately incendiary, practices when pursuing engagement.