Ripples of Real-world Events on Social Media
The rise of social media has drastically lowered the threshold for anyone to communicate in a broad circle of people. In turn, these communication patterns can be leveraged to learn about events from a diversity of perspectives. The talks in this symposium describe different approaches that observe the ripples of Real-world Events on Social Media.
Date: Tuesday 21 March 2017
Time: 12:00 - 15:00
Location: Radboud University, Grotiusgebouw
Room: GR 1.176
12:00 - 12:30
12:30 - 13:00
Nelleke Oostdijk (Radboud University) and Ali Hürriyetoglu (CBS, Radboud University)
Title: Extracting Humanitarian Information from Tweets
In this contribution we present the application of our methods to humanitarian information extraction from tweets and their performance in the scope of the SMERP 2017 Data Challenge task. Detecting and extracting the (scarce) relevant information from tweet collections as precisely, completely, and rapidly as possible is of the utmost importance during natural disasters and other emergency events. We applied a machine learning and a linguistically motivated approach. Both are designed to satisfy the information needs of an expert by allowing experts to define and find the target information. We found that the performance highly depends on the task definition and the ability to facilitate the feedback iteratively. The results of the current data challenge task demonstrate that it is realistic to expect a balanced performance across multiple metrics even under poor conditions.
13:00 - 13:30
Barbara Plank (University of Groningen)
Title: Multi-task learning in NLP: what, how, when?
Multi-task learning (MTL) is quickly gaining popularity in Natural Language Processing (NLP). It has been applied to a range of tasks, from syntactic and semantic sequence prediction, to discourse parsing, and applications in health prediction and machine translation. In this talk I will introduce MTL, how it works and show examples of successful use of MTL. While prospects of MTL are appealing, little is known on *when* MTL works. I will outline recent research in this direction, aiming to shed light on understanding what main-auxiliary task combinations will be effective.
13:30 - 14:00
14:00 - 14:30
Lidwien van de Wijngaert (Radboud University)
Title: Understanding Processes of Structuration on Social Media as a basis for Governmental Communication Strategy
On social media, people (in)directly discuss issues that are related to the government. At the same time governments have difficulties taking a meaningful role in these discussions due to issues of e.g. privacy and (in)equality. This paper aims to obtain insight into the way government can deal with this dilemma. In order to do so, a Dutch Twitter case through qualitative, quantitative and network analysis to obtain insight into the interactions that take place on social media as well as the sources of structure and outcomes related to these interactions. (Adaptive) Structuration Theory is used as a theoretical basis. Cognitive dissonance, Framing and Social Capital provide the theoretical lenses to understand the online interactions. Results show that interactions take place in subgroups that discuss the issue from different angles. Based on these insights, we develop a strategy that helps governments to connect to the public.
14:30 - 15:00
Walter Daelemans (University of Antwerp)
Title: Text analysis for social media moderation
I will report on the text analysis results of the social media moderation application of the AMiCA project. AMiCA (www.amicaproject.be) combined data mining, text and audio analysis, and image processing to detect dangerous content for children and young people in social media. Case studies were the detection of suicide announcements, sexually transgressive behavior, and cyberbullying.
Antal van den Bosch, Louis Onrust, Ali Hürriyetoglu & Florian Kunneman