PHEME Veracity: The 4 th Challenge of Big Data

Transcription

1 PHEME Veracity: The 4 th Challenge of Big Data Tomás Pariente

2 Phemes & social media Memes are thematic motifs that spread through social media in ways analogous to genetic traits We coined the term phemes to add truthfulness and deception to the mix PHEME focuses on a fourth crucial, but hitherto largely unstudied, challenge: Veracity 2

3 Rumour analysis: The Problem Now mostly manual Rumours are challenging Some rumours could take hours, days, weeks or even months to die out Ill-meaning humans can currently outsmart computers (and humans) and appear genuine

4 Rumour analysis: The Problem Mike Brown shot by police in Ferguson We have different rumors emerging from the topic We don t know if they are true. We see the spikes and sometimes they come back (different temporal dynamics) We need to understand the overall conversation to see the different points of view and how the rumours go forward

5 Social Media is Rife with Phemes

6 Social Media is Rife with Phemes

7 From manual to automatic We are investigating... Ontologies for modelling phemes Use a priori knowledge (LOD) and reasoning to detect contradictions Model phemes spread across media, social networks, and time Conversational analysis Real-time rumour classification Pheme visualisation to support veracity checking: media maps, impact maps, geographical maps

8 Technology Outcome: Open Source Computational Framework PHEME VERACITY INTELLIGENCE FRAMEWORK USE CASES Linked Open Data Rumour Ontologies & Reasoning (GraphDB) PatientsLikeMe Cross-Media Content Linking, Spatio-Temporal Grounding Multilingual LOD-Based IE and Opinion Mining Rumour Detection And Veracity Classification PHEME Visual Analytics Dashboard Veracity Intelligence In Patient Care Social Context Models... Historical Data Archive Trust, Authority, Implicit Networks Digital Journalism

9 PHEME Big Data Architecture for veracity analysis Data Value Chain System Workflow Orchestration Multilingual Data Multilingual Social Media Data Multilingual Data Data Messaging / Comms Veracity and Language Value Chain Lang Detection Data Collection IT Big Data Layer Curation Batch Processing Raw data Repository NLP Processing Annotation & Training Processing Event Detection LT Processing & Analytics Storage Infrastructure Cross-media linking Rumour Classification Cross-lingual analysis Usage Stream Processing OntoText GraphDB Knowledge Base Data SW Resource Management Multilingual End Users Data Multilingual Data Pheme Dashboard, Journalist Dashboard IT Value Chain Physical Infrastructure and Virtualization PHEME Some Meeting, Some Place, Some Date

10 Application areas Open-source social intelligence tools for data journalism Involves journalists from SwissInfo.ch, the Guardian, New York Times, and other media Improving healthcare What health-related rumours are discussed in patientclinician consultations Preventative medical advice, e.g. warn patients not to trust certain rumours, when researching their disease online

11 PHEME Dashboard And dynamics Over Time/Location vs replies 11

12 Journalism Dashboard Prototype 12

13 Acknowledgement The PHEME research project has received funding from the European Union's Seventh Framework Programme for research, technological development and demonstration under grant agreement No Thanks! This document does not represent the opinion of the European Community, and the European Community is not responsible for any use that might be made of its content 13