1. DIVE into the Event-Based Browsing of Linked Historical Media
Victor de Boer et al.
Netherlands Institute for Sound and Vision, Hilversum, the Netherlands
Dept. of Computer Science, VU University Amsterdam, the Netherlands
Dutch National Library, the Hague, the Netherlands
Frontwise, Utrecht, the Netherlands

2. Data Sources
NISV: Dutch broadcasting content, including television and radio content. A subset of the NISV collection, videos of news broadcasts, was made available.
KB: The ANP Radio News Bulletin datasets. This dataset is made up of digitized typescripts(radio news scripts, to be read during news broadcasts) from

3. Data Conversion and Enrichment
Extraction
Crowdsourcing
The textual description and descriptive metadata are converted to RDF.
Named Entity Recognition.
Event extraction tools: xTas and Opener.
Crowdsourcing is employed to have human-recognized entities and to refine the results. (CrowdTruth platform)

4. Data Conversion and Enrichment
The data is modelled using the Simple Event Model(SEM). This model allows for the representation of events, actors, locations and temporal description. Am
NISV KB
Amalgame alignment tool
Wikipedia
DBPedia
External sources

5. Demonstration

6. Conclusion
Core contribution is the innovative user interface supporting information interpretation in multimedia collections through dynamic browsing experience with linked data and explicit event representation.

7. Thank you.