Task 3 Web Community Sensing REACTION Mee<ng January 13th, 2012
Team Gustavo Laboreiro - PhD Student @ LIACC, with SAPO Labs grant Jorge Teixeira - SDE @ SAPO Labs, PhD Student @ LIACC Eduarda and Eugénio @ FEUP and Luís @ SAPO New hires: Arian Pasquali 6 month SAPO Labs grant, started February 2012 Nuno Baldaia 8 month REACTION grant, started April 2012 Former collaborator: Matko Bosnjak (Francisca Teixeira, 4 month REACTION grant, starmng June 2012) Future hires: 15 months BI (MSc) @ 980 Eur 8 months BI (BSc) @ 745 Eur
Progress Report Presented TwiVerEcho paper at WWW 2012 Robust language detecmon on TwiVer streams paper (re- submived to CIKM 2012) Experiments with topic- based influence detecmon algorithms (technical report due soon) TwiVerEcho II crawler architecture migramon to big data technologies improved crawl scheduling algorithms
Progress Report InformaMon visualizamon components for REACTION Deployed prototypes at SAPO.pt MVDi: verbetes co- occurrences + interacmve network visualizamon TwiVeuro: Euro 2012 twiver crawl + menmons + senmment analysis + trends visualizamon
MVDi - VisualizaMon hvp://voxx.sapo.pt/mvdi Challenges Web based interacmve network visualizamon Support modern web browsers Client- side vs. server- side layout generamon (performance and layout quality vs. user interacmon and experience)
MVDi - VisualizaMon How it was implemented Using Web standard technologies (HTML, CSS, SVG, JavaScript) On top of open- source JavaScript frameworks (D3.js and jquery) SVG based visualizamons to improve user interacmon and allow custom designs Client- side layout generamon Server data access through the JSONP de facto standard method Problems Performance (SVG does not scale for larger networks) Layout (force layout is not perfect but rather a client- side generated feasible one)
MVDi - VisualizaMon Future improvements BeVer layouts (probably server- side generated) BeVer Mme navigamon interacmon design and larger Mmespan selecmons BeVer zooming features (revealing/hiding details on zooming in/out) Global networks for a given Mmespan (instead of ego networks only) BeVer network exploramon support with new features such as expandable (or commutamve) networks (opening new networks without loosing previous ones)
TwiVeuro Data CollecMon Real Mme collecmon of tweet with the #Euro2012 hashtag Using TwiVer Streaming API 51 tweets/sec limit (e.g., reach when there is a goal) 3 servers collecmng data for redundancy
TwiVeuro Real- Mme data collecmon and storage InformaMon extracmon SenMment analysis InformaMon visualizamon
TwiVeuro Data Storage +5,5 million tweets stored since 8 june and indexed in Solr IniMally, Up to 30 sec between user published a tweet unml it got indexed in our system Currently, 1 sec on average
TwiVeuro - VisualizaMon hvp://twiveuro.sapo.pt Challenges Support modern browsers and mobile devices Access and display data changes on a near realmme basis Encourage user parmcipamon
TwiVeuro - VisualizaMon How it was implemented Using Web standard technologies (HTML, CSS, SVG, JavaScript) On top of open- source JavaScript frameworks (D3.js and jquery) and an open- source responsive CSS framework to reduce the design and development Mme (TwiVer Bootstrap) SVG based visualizamons to improve the user interacmon and allow custom designs Server- side data access through the JSONP de facto standard method to load and refresh data
TwiVeuro - VisualizaMon Problems Performance on mobile devises (parmally solved by disabling animamons on such devices) Adaptability of SVG based visualizamons to mobile device sizes Visualizing real Mme changes Future improvements BeVer overall performance BeVer Mme navigamon interacmon design SenMment visualizamons New visualizamon tools to support post event research
TwiVeuro MenMons ExtracMon Complex name disambiguamon Popular names and surnames MulMple ways of referring to the enmmes
Complex name disambiguamon Progress Report Eduardo Alves da Silva (CroaMa) Bruno Eduardo Regufe Alves (Portugal) Eduardo dos Reis Carvalho (Portugal) David Josué Jiménez Silva (Spain) Fábio Alexandre da Silva Coentrão (Portugal) João Pedro da Silva Pereira (Portugal) David Forde (Republic of Ireland) David Lafata (Czech Republic) David Limberský (Czech Republic) Hugo Miguel Pereira de Almeida (Portugal) João Filipe Iria Santos MouMnho (Portugal) CrisMano Ronaldo dos Santos Aveiro (Portugal) Rui Pedro dos Santos Patrício (Portugal)
Popular names and surnames ChrisMan Ulf Wilhelmsson (Sweden) ChrisMan Bjørnshøj Poulsen (Denmark) ChrisMan Dannemann Eriksen (Denmark) ChrisMan Maggio (Italy) Tomáš Hübschman (Czech Republic) Tomáš Pekhart (Czech Republic) Tomáš Necid (Czech Republic) Tomáš Sivok (Czech Republic) ChrisMan Bjørnshøj Poulsen (Denmark) Jakob Bendix Uhd Poulsen (Denmark) Simon Busk Poulsen (Denmark)
MulMple menmons Itália Italy ITA it Italians Italian (team) Italianos... Gli Azzurri The Azures CrisMano Ronaldo CR7 PT07 Portugal7 POR07...
Matches and opinions portugal- alemanha POR- DEN nl- pt cze- pt... like dislike +++ - - - - - - fail lousy amazing :- (...