Social Mining Big Data Ecosystem H 2020 www
Social Mining & Big Data Ecosystem – H 2020 www. sobigdata. eu Sentiment-enhanced Multidimensional Analysis of Online Social Networks: Perception of the Mediterranean Refugees Crisis Mauro Coletto∗†, Andrea Esuli†, Claudio Lucchese†, Cristina Ioana Muntean†, Franco Maria Nardini†, Raffaele Perego†, Chiara Renso† ∗ IMT School for Advanced Studies Lucca - ITALY † ISTI - CNR Pisa - ITALY
Refugee Migration in Europe 1 MILLION REFUGEES IN 2015 (source UNHCR)
Analytical Framework • An analytical framework to interpret social trends from large tweet collections by extracting and crossing information about the following three dimensions: – Time – Space (user and mentioned locations) – Sentiment (user sentiment and tweet sentiment) • Refugee crisis discussions over Twitter as the case study (5. 08. 2015 – 17. 09. 2015) – – How is the European population perceiving this phenomenon? What is the general opinion of each country? How is perception influenced by events? How does perception evolve in time in different European countries?
Dataset Major events Tweets per country
Dataset Major events Relevant English tweets per day
Methodology 1. 2. Extract temporal information at per day level Extract relevant spatial information: Ø Ø 3. Enrich data with sentiment information Ø Ø 4. User location Location mention Tweet sentiment User sentiment Perform multidimensional analyses considering content and locations in time
Deriving Sentiment Initial seeds #refugeeswelcome Positive Hashtags Negative Hashtags #refugessnotwelcome Enrich hashtag seeds from #-tag co-occurrence Positive Users Negative Users Positive Tweets Negative Tweets
Deriving Sentiment M. Coletto, C. Lucchese, S. Orlando, and R. Perego, “Polarized user and topic tracking in Twitter, ” in SIGIR 2016, Pisa, Italy, 2016.
Deriving Sentiment Pro refugees • • • • #campliberty #health #humanrights #marchofhope #migrantmarch #refugeecrisis #refugeemarch #refugeescrisis #sharehumanity #solidarity #syriacrisis #trainofhope Against refugees • • • • • #illegalimmigration #illegals #invasion #isis #islamicstate #justice #migrantcrisis #muslimcrimes #muslims #no 2 eu #nomoremigrants #nomorerefugees #patriot #security #stoptheeu #taliban #terrorism
Refugees Crisis Perception Analysis • AQ 1: What is the evolution of the discussions about refugees migration in Twitter? • AQ 2: What is the sentiment of users across Europe in relation to the refugee crisis? What is the evolution of the perception in countries affected by the phenomenon? • AQ 3: Are users more polarized in countries most impacted by the migration flow?
Space and Time analysis European country mentions AT-HU border opens Refugess at Macedonia border Africa & Middle East country mentions News about Syria Flow shift to Croatia Terrorist attack in Nigeria
Sentiment Analysis • Internal and external perception by country – Index ρ - the ratio between pro refugees users and against refugees users – Red means a higher predominance of positive sentiment, higher ρ – Yellow means a higher predominance of negative sentiment, lower ρ - + (a) Overall. - + (b) Internal perception. - (c) External perception. +
Sentiment Analysis in UK • Positive and negative users for different cities in UK before and after September 4 (death of Alan Kurdi, borders between AT HU, Germany welcomes refugees). – bars show the number of polarized positive and negative users by city – the heat map in background indicates the value of ρ
Sentiment Analysis • For mentioned locations analysis and tweet sentiment Hungary Croatia Germany
Conclusions and Future Work • Adaptive and scalable multidimensional framework to analyze the spatial, temporal and sentiment aspects of a polarized topic discussed in an online social network. • The combination of the sentiment aspects with the temporal and spatial dimension is an added value that allows us to infer interesting insights. – The analyzed European users are sensitive to major events and mostly express positive sentiments for the refugees, but in some cases this attitude suddenly changes when countries are exposed more closely to the migration flow. • Future work: – real-time streaming scenario – analyze the network relationships in the Twitter user graph – multilingual analysis
Thank you • Questions? • Contact: cristina. muntean@isti. cnr. it
- Slides: 16
