Semantic media mining: what? why? how?

[2014-09-05 Fri 19:07] speaker: Gerard Dupont, Airbus Defence and Space event: Friday Hacks, NUS Hackers ** Airbus defense and space rockets, satellites, defence systems for communication intelligence and security ** data processing of public data sources ** activities that cannot be automated

• expert analysis of content
• link and mapping heterogenous information
• reliability and assessing information
• report and synthesis of information ** processing issue need a combination of tools to address specific intelligence needs ** web as a mean for people safety
• tweets move faster than the earthquake (NZ)
• NEA interested in information on litter on the ground
• SCDF interested in information of fire, accidents ** Open Source INTtelligence needs
• credibility evaluation similar information from other sources
• finding the original source
• detecting false information
• fusing facts and events ** Open Source INTtelligence challenges
• limited processing resources
• air gap between aquisition and exploitation, public vs secure network in the military
• private data management private data is any piece of information that can be linked to a person ** WebLab platform team started out in text-mining of Airbus planes documentation data to find relevant information for maintenance

integration of tools (Tika, GeoNames, etc) for collections, processing, analysis, communication of open source information. Able to switch tools easily for different customers. ** WebLab architecture *** core SOA infracstructure for unstructured information processing components *** services COTS, open source, advance research components *** applications ** technology platform SOAP service integration, RDFS knowledge model process one document at a time

specific tools

• Tomcat
• Apache Camel
• Apache Jackrabbit
• GeoServer ** semantic matches alert on helicopter “Eurocopter delivers its 1000th Dauphin to Pawan Hans in India” semantic extractor recognized Dauphin is the AS365 helicopter ** knowledge base of filtered and validated information ** WebLab the open source project hosted in OW2 weblab.ow2.org ** applications *** application for Green Peace video -> audio -> cleaned audio -> japanese text -> english test -> enriched text

Raise alert if keywords are detected. To reduce the number of documents an analyst needs to look at. *** index news videos to make them searchable *** network of terrorists