towards comprehensive community based disaster management
TRANSCRIPT
Social Computing and Crisis in the New Information Age
Marc van den Homberg, 8 th of October 2013
Topics HFM -RWS 241 “Social Media and Information Technology for Disaster and Crisis Response”
Current practices in information sharing
Novel technologies for information sharing and disaster information
coordination
Social trust, social cyber-threats, and related socio-culture aspects
Development of an exercise in information sharing in a disaster
scenario.
Current practices in information sharing
Source: UN OCHA
Humanitarianism in the network age, 2013,
Nethope report 2010
New practices…
Traditional practices
of information
sharing
New practices, social media
Challenges of information sharing betweenaffected and supporting community
Multitude of data sources
Coupling of traditional with social media data
Aggregation and validation of data
Information not well adapted to local context
(illiteracy, language, documented expertise vs storytelling)
Technology gap
Using too advanced technology which does not ‘reach’ communities
Often one-way communication, with a very limited feedback loop
Collaboration gaps
Reluctance between actors to share information
Professionals versus volunteer communities
Novel technologies for information sharing and disaster information coordination
Community based comprehensive recovery
www.cobacore.eu
Collaboration gaps right after a disaster
respondingprofessionals
respondingcommunity
affected community
resilient community
From external response to community driven crisis response
individual and collective needs
individual and collective capabilities
Current response models mostly between affected community and responding professionals
COBACORE includes the interaction with responding community
10
COBACORE; EU FP7 project No. 313308
Expert finder
Pluralism monitor
Media miningPress freedom
ZimbabweGlobal
Expert panel for validation (media experts, a.o. IPDC UNESCO)
Data acquisition
Annotation Analysis Interpretation
TOPICS ACTORS
AT
TE
NT
ION
� Topics (e.g., agriculture, politics,
economy)
� Election news or not
� Foreign news versus national and
local news
� Specific events
� Regions
Individuals
� names
� affiliations (with groups in society)
� gender of individuals
� pro or against government
� function (MP, minister, ..)
� political party
� government versus opposition
Groups
� religious
� ethnic
� NGO
� political party
SE
NT
IME
NT Tone (positive/negative)
Use of hate speech
Annotation
Incremental supervised learning architecture
Text analysis
engine
Topic
detection/
classification
Sentiment
analysis
Named entity
recognition
Online news
incremental learning
supervision
DB/KB
Demo: text mining on news data in action!
Twitcident
Filtering social media (tweets) to obtain real-time intelligence for support of operational emergency services. Also for prevention and analysis.
Contact details
Marc van den Homberg T +31 88 8667135
Senior business consultant ICT4D M +31 6 51069884
Oude Waalsdorperweg 63,
2597 AK Den Haag E [email protected]
The Netherlands S marcvandenhomberg