News Innovation Lightning Talk

Editor's Notes

• #3: Linked Data is the friendly face of the semantic web. Its a community effort to (largely) ditch the wider vision of the semantic web and instead focus on pragmatic results – like actually publishing some data on the web Linked Data involves putting data online, in a very fine-grained way. So we don't just get cool URIs for datasets, but also the concepts and resources they refer to: people, places, things. The Who, Why, What, Where and When. Direct integration of data with the web by using URLs to identify stuff.
• #4: I'll note that the fundamentals of the semantic web technology is actually very simple. Simple enough that I could teach it to my six year old. He's become something of a domain expert on Star Wars. Here's our attempt to write some stuff down to help him figure a few things out. Turns out he was better at it than me. And he was able to learn something new.
• #5: Linked Data effort has been successful. Large amount of data become available in just a few years. Some high quality, some low. But thats the nature of the web. The volume of data is growing and there's real momentum behind it now from various organisations including the BBC and the New York Times.
• #6: So Linked Data can be another tool to help people tell stories with data. By presenting inter-linked facts and figures, it could start to help people make sense of stuff and add narrative to it, to explain it to others. What's interesting is that it has the potential to create some new and innovative ways to explore datasets. But thats not the area I want to focus on today. There are two aspects that I want to draw out.
• #7: The first is Context. If you're telling a story with some data, then you're contextualising it with a narrative. But we really need to preserve the relationship between that narrative and the original source – we need to link them together. Because then we can drill down to the data and explore it ourselves. The data can also be live. With linking between data and stories, we can also find other stories that use the same data. About the same place, or thing. And we can do that in a very fine-grained way. Makes for interesting ways to explore other analyses, other viewpoints. Navigate a web of stories, the concepts to which they refer to, the people telling them, etc. Thats a rich fabric
• #8: The second is Provenance By preserving links to data, we can back-track to the source. This means we have the potential to do things like fact-checking. We can start to discover the reliability, and accuracy of the data. But we can also look outward, and then find all the stories that might be invalidated because of data errors or quality issues. We can find those stories that may be drawing on questionable sources to draw inaccurate conclusions. Potentially even find derived data and trace from there. Previously thats been hard to do. And I don't see other technologies looking at that angle, or those that could possibly work without using fine-grained linking and a common model – thats what RDF and linked data is.
• #9: So whether we want to capture and publish all facts and figures using RDF is a separate issue to whether its useful to begin connecting up sources and stories using Linked Data. Convergence happens naturally as data starts to be opened up. We can share identifiers for things and their relationships. Its not just possible or desirable for that it happen. It is happening. There's evidence for it on the diagram. The areas with the same colour are same kind of data. They share/reuse concepts The links show interconnections and share identifiers. Thats what dbpedia is. Its becoming a linking hub for the semantic web. A place for accumulating shared ids.