Guest Lectures 2017/18


October 2017

10/09, 2-4 pm, LK 056

Hanna Krasnova, University of Podsdam: Social Media and Mobile: The Bright, the Dark and the Ugly
10/30, 2-4 pm, LC 126 Mykola Pechenitzkiy, TU Eindhoven: Responsible Learning Analytics: Facets of (un)fairness and (non)transparency
November 2017
11/20, 2-4 pm, LC 126 Michael Emmerich, Leiden University: Simulation of complex networks
11/27, 2-4 pm, LC 126 Wil van der Aalst, TU Eindhoven: Process Mining: Data Science in Action
December 2017
12/04, 2-4 pm, LC 126 Sonja Utz, IWM Tübingen: Effects of social media use: what we have learned from the ReDefTie project
12/18, 2-4 pm, LC 126 Martijn Willemsen, TU Eindhoven: Understanding user preferences and goals in recommender systems
January 2018
01/15, 2-4 pm, LC 126 Krishna Gummadi, Max Planck Institute for Software Systems: Privacy, Fairness, Transparency, and Control of Targeted Advertising on Social Media
 29/01, 2-4 pm, LC 126 Inge Molenaar, Radboud University: Measurement and support of self- and social regulated learning in advanced learning technologies
February 2018
02/06, 2-4 pm, LC 126 Rodney Clarke, Wollongong Univeristy: A Visualisation of a Semantic Social Media Sentiment Analysis

Lecture slides (available for RTG members only)

Social Media and Mobile: The Bright, the Dark and the Ugly
Social Media platforms transform the society we live in today: they change the way we present ourselves, communicate with each other and spend our free time. As these platforms are increasingly permeating our daily routines, conversations and leisure, many keep asking themselves about the meaning and long-term consequences of these changes. On the one hand, proponents express strong optimism in their positive value, viewing Social Media platforms, such as Facebook, as a source of significant benefits at both individual and social levels. On the other hand, opponents express deep concerns over the dangers strong engagement with Social Media brings along. Critics link participation on Social Media to privacy threats such as information collection, aggregation, as well as stalking and bullying. However, despite the presence of these imminent threats, users readily reveal striking amounts of information in their communication online. Bewildered by these dynamics, many wonder about the rationale behind user sharing behavior. Are users at all concerned about their privacy and if so, to what extent? However, these are not only privacy threats users have to put up with. A number of most recent reports have linked the use of Facebook to feelings of inadequacy, loneliness, depression, and an array of other psychological deviations. Evidently, these negative outcomes cannot be solely linked to the privacy concerns users experience. Our research suggests that outcomes of SNS use are a function of usage patterns, with research differentiating between passive and active components of usage. Following these findings, we uncover the proliferation of upward social comparison and envy feelings among SNS members as a result of passive following on Facebook. This finding is of tremendous social importance, as it may help to explain an array of negative emotional outcomes linked to SNS use. Finally, the obsession with the mobile devices as such has recently become obvious. We observe individuals forgo face-to-face communication, parental responsibilities and romantic engagements in favour of smartphone-enabled communication. Especially, parental addiction emerges as a troublesome development, with many parents neglecting their children and thereby threatening their cognitive and physical developments, and above all the safety of their offsprings. Most recent findings in this area will be presented during the talk.
Responsible Learning Analytics: Facets of (un)fairness and (non)transparency

Application-driven research in predictive analytics contributes to the massive automation of the data-driven decision making and decision support. Many of these decisions affect our everyday life and its future. Data mining researchers and practitioners often have a (false) believe that data mining techniques have no bad intents. In this talk I will revisit several popular applications of predictive analytics to highlight why the general public, domain experts and policy makers have good reasons to consider off-the-shelf tools as a thread. In particular, it becomes better understood that predictive models may systematically discriminate groups of people even if data mining researchers and practitioners have only good intentions when they develop and apply predictive analytics. I will present different facets of discrimination-awareness and transparency in analytics and reflect on the current state-of-the-art and further research needed for gaining a deeper understanding of what it means for predictive analytics to be ethics-aware, transparent and accountable.
Simulation of Complex Networks
The lecture deals with efficient algorithms for network simulation.
The simulation of complex networks involves the random generation of graphs with predefined properties, which serve as null or ideal models in the investigation of real networks, as well as the simulation of dynamic processes due to given behavioral patterns of the network. Applications arise in systems research (e.g., network medicine), epidemiology, and social network analysis (e.g., information dissemination).
Process Mining: Data Science in Action
Process mining provides new ways to utilize the abundance of event data in our society. This emerging scientific discipline can be viewed as a bridge between data science and process science: It is both data-driven and process-centric. Process mining provides a novel set of tools to discover the real process, to detect deviations from normative processes, and to analyze bottlenecks and waste. Analogous to spreadsheets, process mining provides a generic domain-independent technology (starting from events rather than numbers). In his talk, Wil van der Aalst will argue that process mining should be an integral part of tomorrow's data scientist. He will introduce basic concepts, explain a particular discovery technique (inductive process mining), and elaborate on his collaboration with industry. His research group at TU/e applied process mining in over 150 organizations, developed the open-source tool ProM, and influenced the 20+ commercial process mining tools available today. 


Effects of social media use: what we have learned from the ReDefTie project
Sonja Utz

Using social media has become part of the daily routine for many people, and many scholars have examined how social media use affects people in the last years. However, many of these studies were cross-sectional, relied on student or other convenience samples and/or focused on the effects of Facebook (or its national predecessors). In this talk, an overview over the main results from the ERC project “Redefining tie strength: How social media (can) help us to get non-redundant useful information and emotional support” (REDEFTIE) will be given. In this project, experimental studies were accompanied by a longitudinal study with a representative sample of Dutch online users. The project focused not only on Facebook use, but also on the use of professional networks such as LinkedIn or microblogging services such as Twitter, and it did  not only examine effects on indicators of well-being, but also at informational benefits at work or civic engagement. The talk will cover short- and long-term effects of Facebook use on emotions and well-being, as well as work on the effects of LinkedIn use on informational benefits and the role of ambient awareness. Despite the limitations of specific methods, together, the studies give a comprehensive picture over the effects of social media use in these domains.

 Understanding user preferences and goals in recommender systems
Martijn Willemsen
Recommender systems typically use collaborative filtering: information from your preferences (i.e. your ratings) is combined with that of other users to predict what other items you might also like. Much of the research in the field has focused on building algorithms that provide recommendations based purely on predicted accuracy. However, these models make strong assumptions about how preferences come about, how stable they are, and how they can be measured. Having a background in decision psychology I have studied how the preference elicitation method of recommender systems can be better understood and improved based on psychological insights. I will illustrate this with an example of new choice-based preference interfaces we have developed.
Moreover, recommender systems should also align with user goals. Many real-life recommender systems are evaluated mostly on (implicit) behavioral data such as clicks streams and viewing times. However, such an approach has limitations and I will show how a user-centric approach can help better understand why users are satisfied or not, for example why users prefer diversification over prediction accuracy as it reduces choice difficulty. The behaviorist approach to evaluation also misses that users’ short term goals (i.e. their current behavior) might not be representative of the goals they want to attain (i.e. their desired behavior). This is especially relevant in health and life style domains where people are in need of support while changing their current behavior. I will elaborate on an example in the energy recommendation domain, and show how a different type of recommender approach and interface might help users to save more energy.
Privacy, Fairness, Transparency, and Control of Targeted Advertising on Social Media
Krishna Gummadi

All popular social media sites like Facebook, Twitter, and Pinterest are funded by advertising, and the detailed user data that these sites collect about their users make them attractive platforms for advertisers. In this talk, I will first present an overview of how social media sites enable advertisers to target their users. Next, I will pose and attempt to answer the following four high-level questions related to privacy, fairness, transparency and control of social media advertising today.

  1. Privacy threats: what personal information about users are the sites leaking to advertisers to enable targeted ads? 
  2. Fairness: can an advertiser target users in a discriminatory manner? If so, how can we detect and prevent discriminatory advertising?
  3. Transparency: can users learn what personal data about them is being used when they are targeted with an ad?
  4. Control: can users control what personal data about them is being used when they are targeted with an ad?
Measurement and support of self- and social regulated learning in advanced learning technologies
Inge Molenaar
The increased use of advanced learning technologies (ALTs) is not only an opportunity for supporting learning, but also new source for data collection. Online (multiple) data stream(s) provides a fundamentally new approach to the measurement of S(S)RL during learning. The goal of this lecture to discuss:
a) How we gather and analyze trace data to measure students' SRL during learning? and b) How these measurements can be used to support learners’ self- and socially regulated learning in ALTs? This is the focus of research in the adaptive learning lab (ALL) and of our international collaborators in the Earli Center of Innovative Research. First, I will discuss the theoretical background and current empirical status around the measurement and support of S(S)RL in ALTs. Second, I will elaborate on two projects we are currently working on in the adaptive learning lab.
a. The VENI project investigates variation in children’s’ effort and accuracy while learning math in an adaptive learning technology. Students learning in ALTs on tablets leave rich traces of data that capture many details of their learning process (Gašević et al., 2015). Although ALTs successfully use student data to adjust instruction to learners performance, they fail to use captured data to support self-regulated learning (Winne & Baker, 2015). We are exploring what “moment-by-moment learning curves” (Bakker et al. 2103) reveal about students’ self-regulated learning and whether these curves can be used as personalized dashboard to guide young learners regulation.
b. The STULE project aims to support secondary vocational students in task oriented reading and the regulation thereof. Students work in groups and their reciprocal peer tutoring is guided by a macro-script. The script focusses on eliciting group discussions around task perception and the selection of appropriated reading strategies. Currently we are in the processes of developing a classifier to automatically detect the students’ strategy use during reading.

A Visualisation of a Semantic Social Media Sentiment Analysis
Rodney Clarke

Sentiment Analyses are widely used approaches to understand and identify emotions, feelings, and opinion on social media platforms. Most sentiment analysis systems measure the presumed emotional polarity of texts. While this is sufficient for some applications, these approaches are very limiting when it comes to understand how social media users actually use language resources to make sense of extreme events. In this paper, the authors apply a Sentiment Analysis based on the Appraisal System from the theory of communication called Systemic Functional Linguistics to understand the sentiment of event driven social media communication. A prototype was developed to code and visualise geotagged Twitter data using the Appraisal System. This prototype was applied to tweets collected during and after the Sydney Siege, a hostage situation in a busy café in Sydney’s inner city at the 15th of December 2014. Because the Appraisal System is a theorised functional communication method, the results of this analysis are more nuanced than is possible with traditional polarity based sentiment analysis.