Work

Publications

The Effect of Collective Attention on Controversial Debates on Social Media
To appear at WebSci 2017.

We study the evolution of long-lived controversial debates as manifested on Twitter from 2011 to 2016. Specifically, we explore how the structure of interactions and content of discussion varies with the level of collective attention, as evidenced by the number of users discussing a topic. Spikes in the volume of users typically correspond to external events that increase the public attention on the topic, for instance, discussions about `gun control’ often erupt after a mass shooting. We find consistent evidence that  increased collective attention is associated with increased network polarization and network concentration within each side of the debate.
Collaborators: K. Garimella, G. De Francisci Morales, A. Gionis

Where could we go? Recommendations for groups in location-based social networks.
To appear at WebSci 2017.
We consider the problem of recommending a list of POIs to a group of users in the areas that the group frequents. Our data consist of activity on Swarm, a social networking app by Foursquare, and our results demonstrate that Geo-Group-Recommender (GGR), a hybrid recommender system that combines collaborative filtering, content features and kernel density estimation for geographical personalization outperforms a large number of other models. Moreover, we find evidence that user preferences differ both venue category and location between individual and group activity. We also show that combining individual recommendations using group aggregation strategies is not as good as building a profile for a group.
Collaborators: Frederick Ayala, B. Daróczy, A. Benczur, A. Gionis

Factors in Recommending Contrarian Content on Social Media.
To appear at WebSci 2017.
We propose an algorithmic solution to the problem of reducing polarization. The core idea is to expose users to content that challenges their point of view, with the hope broadening their perspective, and thus reduce polarity. Our method takes into account several aspects of the problem, such as the estimated polarity of the user, the probability of accepting the recommendation, the polarity of the content, and popularity of the content being recommended.

Ad-blocking: A study on Performance, Privacy, and Counter-measures.
To appear at WebSci 2017.
we study the performance of popular ad-blockers on a large set of news websites. Moreover, we investigate the bene ts of ad-blockers on user privacy as well as the mechanisms used by websites to counter them. Finally, we explore the traffic overhead due to the ad-blockers themselves.
Collaborators: K. Garimella, O. Kostakis

Tutorial: Polarization on Social Media.
To appear at AAAI ICWSM 2017.
This tutorial presents a systematic review of the study of polarization as manifested online, in particular on social media. In the first part, we clarify the concept of polarization, and the related nomenclature, by drawing from the social and political sciences. Then, we focus on the understanding of polarization as manifested online. In the second part, we review techniques for detection, quantification, and mitigation of polarization.
We conclude the tutorial by presenting open challenges and promising research directions for researchers interested in this area.
Collaborators: K. Garimella, G. De Francisci Morales, A. Gionis

The ebb and flow of polarized debates online.
To appear AAAI ICWSM 2017. [Slides]
We study polarized debates on Twitter from 2011 to 2016. Specifically, we explore how polarization varies (i) over time and (ii) with the level of interest in a topic, as evidenced by the number of users discussing a topic.
Collaborators: K. Garimella, G. De Francisci Morales, A. Gionis

Mary, Mary, Quite Contrary: Exposing Twitter Users to Contrarian News.
Demonstration – appeared at WWW2017.
We visualize polarized discussions on social media. Using our visualization, users can understand how discussions are shaped and explore the positions of the various actors. On top of that, we demonstrate a completely automated pipeline to recommend content to users from the ‘opposite side’, i.e. content that might not agree with their beliefs.
Collaborators: K. Garimella, G. De Francisci Morales, A. Gionis

Reducing Controversy by Connecting Opposing Views.
Appeared at WSDM 2017. [ACM][Slides]
Best Student Paper
When polarized issues emerge on social media, we often observe the creation of ‘echo chambers’. We study algorithmic techniques to bridge these chambers.
Collaborators: K. Garimella, A. Gionis, G. De Francisci Morales
Covered on Acolyer, Aalto News, Communications of ACM – News, Science Newsline Phys.org.

Local Discrepancy Maximization on Graphs.
Invited to TKDE 2017. [IEEEXplore]
Extended version of our ICDE2015 ‘Bump Hunting’ paper.
Collaborators: A. Gionis, A. Ukkonen

Modeling Urban Behavior by Mining Urban Data.
IEEE Transactions on Big Data / Special Issue on Urban Computing. [IEEEXplore]
We present a technique to discover how different regions of cities are associated with different kinds of activity.
Collaborators: E. Çelikten, G. Le Falher

“What Is the City but the People?” Exploring Urban Activity Using Social Web Traces.
Demonstration – appeared at WWW2016. [ACM] [pdf]

GeoTopics is a system to explore geographical patterns of urban activity, based on Foursquare check-ins.
Collaborators: Emre Çelikten, Géraud Le Falher

Quantifying Controversy in Social Media.
Appeared at WSDM 2016. [ACM] [Arxiv] [Slides]
We perform a systematic methodological study of controversy detection using social-media network structure and content. 
Collaborators: K. Garimella, A. Gionis, G.D.F. Morales
Covered on MIT Technology Review, NBC News, and elsewhere.

Exploring Controversy on Twitter.
Demonstration – appeared at CSCW 2016. [ACM] [ Webpage ]
We perform a systematic methodological study of controversy detection using social-media network structure and content. 
Collaborators: K. Garimella, A. Gionis, G.D.F. Morales

Absorbing random-walk centrality: Theory and algorithms.
Appeared at ICDM 2015. [Arxiv] [Slides] [Blog Post]
We study a notion of network centrality based on absorbing random walks. Given a graph G and a set of query nodes Q, we want to identify the k most central nodes in G with respect to Q. In our model, we consider central nodes to be absorbing for random walks that start at the query nodes Q. The goal is to find the set of k central nodes that minimizes the expected length of a random walk until absorption.
Collaborators: A. Gionis, H. Mavroforakis

Bump hunting in the dark: Local discrepancy maximization on graphs.
Appeared at 
ICDE 2015. [IEEEXplore] [PDF] [Slides] [Poster[Code]
We study the problem of discrepancy maximization on graphs: given a set of nodes Q of an underlying graph G, we aim to identify a connected subgraph of G that contains many more nodes from Q than other nodes. This variant of the discrepancy-maximization problem extends the well-known notion of ”bump hunting” in the Euclidean space.
Collaborators: A. Gionis, A. Ukkonen

Similar Neighborhoods Across Cities.
Appeared at ICWSM 2015 and IC2S2. [PDF[Slides] [Blog Post] [Abstract] 
We develop methods to find similar neighborhoods across different cities using data collected from geo-enabled social-media platforms and look for the best features and measures to define similarity of neighborhoods in cities.
Collaborators: G. Le Falher, Aris Gionis

Public transport, urban activity and housing prices.
Appeared at MUD 2015. [PDF]
We study how proximity to urban activity centres affects housing prices in Helsinki. Collaborators: I. Zliobaite, T. Lehtiniemi, P. Parviainen and T. Janhunen.

Detecting Prominent Patterns of Activity in Social Media, PhD Thesis, 2013
Advisor: Nick Koudas, Committee: Vassos Hadzilacos, Peter Marbach, Panos Ipeirotis.

Available on the UofT Library.

Sparsification of influence networks.
Appeared at KDD 2011. [ACM][Code]
We present Spine, an efficient algorithm for finding the “backbone” of an influence network.
Collaborators: Francesco Bonchi, Carlos Castillo, Aristides Gionis, and Antti Ukkonen.

Identifying, attributing and describing spatial bursts.
Appeared at VLDB 2010. [ACM]
We propose algorithms to identify geographically focused information bursts, attribute them to demographic factors and identify sets of descriptive keywords.
Collaborators:
Nilesh Bansal, Nick Koudas.

Twittermonitor: trend detection over the twitter stream.
Appeared at SIGMOD 2010. [ACM]
We present TwitterMonitor, a system that performs trend detection over the Twitter stream. The system identifies emerging topics (i.e. ‘trends’) on Twitter in real time and provides meaningful analytics that synthesize an accurate description of each topic.
Collaborator: Nick Koudas.

Early online identification of attention gathering items in social media.
Appeared at WSDM 2010. [ACM]
We consider the problem of early online identification of items that gather a lot of attention in social media. We present sequential statistical tests that enable early identification of attention gathering items.
Collaborator: Nick Koudas.

Efficient identification of starters and followers in social media.
Appeared at EDBT 2010. [ACM]
We formalize notions of `starters’ and `followers’ in social media and study random sampling approaches allowing us to achieve significant efficiency while identifying starters and followers.
Collaborator: Nick Koudas.

A study on workload-aware wavelet synopses for point and range-sum queries.
Appeared at DOLAP 2006. [ACM]
We perform an extensive theoretical and experimental study on common synopsis construction algorithms, with emphasis on wavelet based techniques, that take under consideration query workload statistics.
Collaborators: Dimitris Sacharidis, Timos Sellis.


Ongoing Work

The following are ongoing pieces of work that have not been published at reviewed venues (yet). If you are interested in knowing more about the projects, feel free to contact me.

Quantifying Controversy on Social Media – Extended. Submitted.
Journal version of the original ‘controversy’ paper.
Collaborators: K. Garimella, G.D.F. Morales, A. Gionis

Lines of Work. Submitted.
We analyze a large citation dataset to discover lines of work, i.e. sets of research papers that represent research areas. The project is a continuation of Karmen’s student project.
Collaborators: K. Dykstra, A. Gionis, N. Tatti, J. Lijffijt

Markov Chain Monitoring.
We study algorithms to effectively monitor traffic on a network under Markov Chain assumptions.
Collaborator: Evimaria Terzi

The biases of Foursquare activity.
We study how the urban activity reported on Foursquare varies for different segments of its user base.
Collaborators: Ian Qihang Gu, Dimitris Sacharidis



List of Publications
also available on Google Scholar, ACM, and DBLP.


Professional Service
Reviewer or committee member for the following venues:

2018: ICDE, WSDM.
2017: SIG KDD, WWW, ICWSM, ICDE, Information Retrieval (INRT), Workshop on Social News On the Web (SNOW), Data Science in Societal Debates (DSSD), IEEE Transactions on Big Data (TBD).
2016: VLDB Journal, WWW, Transactions on Knowledge and Data Engineering (TKDE), Workshop on Social News On the Web (SNOW), Workshop on Social Media and Risk (SoMeRis), Transactions on Intelligent Systems and Technology, Transactions on Information Systems, ECML PKDD Journal (DAMI).
2015
: VLDB Journal, ECML PKDD Journal (DAMI), CIKM, WWW, Internet Mathematics, Transactions on Information Systems (TOIS), Knowledge and Information Systems (KAIS).
2014: WWW, CIKM, ICWSM, Transactions on Parallel & Distributed Systems (TPDS), Transactions on Dependable and Secure Computing (TDSC).

Talks
Intl. Workshop on Machine Learning for Large Scale Networks, Alaska, USA, May 2017,  ‘Polarization on Social Media’
Lyon I, Claude Bernard, May 2017, ‘Social Media Analysis’
INSA Lyon, May 2017, , ‘Social Media Analysis’
NYU Abu Dhabi, April 2017, ‘Polarization on Social Media’
London School of Economics, March 2017, ‘Polarization on Social Media’
Elisa, Helsinki, Finland, Dec. 2016, ‘Urban Computing’
Futurice, Dec.2016, ‘Expert Finding in Social Networks’
PyCon Finland, Nov. 2016, ‘Visualizing Twitter Discussions with NetworkX’
Sanoma, Dec. 2015, ‘Social Media Analysis’
Inria, Lille, France, May 2015, ‘Absorbing Random Walk Centrality’


Teaching

Modern Database Systems, Spring 2017, with Aristides Gionis.
Modern Database Systems, Spring 2016, with Aristides Gionis.
The course covers (i) relational database systems and disk access analysis, (ii) querying and indexing for semi-structured data and text, (iii) map-reduce and spark.
Mining the Social Web, Spring 2015, with Aristides Gionis.


Advising
Served / serving as advisor for:

  • Jere Vaara’s BSc thesis (2016).
    Topic: ‘A Facebook bot for schedule coordination’.
  • Miika Rantakaulio’s BSc thesis (2016).
    Topic: ‘A Facebook bot for place recommendation’.
  • Clemens Westrup’s MSc thesis (2016), supported by Sanoma.
    Title: ‘An Exploration of Representation Learning and Sequential Modeling Approaches for Supervised Topic Classification in Job Advertisements’.
  • Julien Blegean’s MSc thesis (2015).
    Title: ‘Twitter the Rioter : Analyzing roles through a protest on social media. What was your part during the 2014 Ferguson riots?’
  • Geraud Le Falher’s MSc thesis (2014).
    Title: ‘Finding similar neighborhoods across cities by mining human urban activity’.