I’ve had my doubts about the significance of context from time to time, but certainly in certain, er, contexts, it is pretty much essential. Case in point, news, as Chris Saad talks about in Context is Everything. News Should Be A Collaborative Work. Chris argues that “News is broken.”, and I don’t disagree, but would add the argument that news is just a special case of information in general. Take Chris’ list of what needs to be added to news stories (through a link at least):
- Timeline of events for this story (circa does an amazing job here)
- Broader historical context and precedents
- Supporting/contradicting facts (preferably in charts and graphs)
- A comprehensive and CURATED compilation of related coverage and conversation in other places,
- Biases of the reporter and contributing sources
- Biases of the subjects of the story (particularly if politicians are involved)
This is all (potentially) Linked Data that can radically enhance the information provided by the news article.
But take a different piece of information, say data associated with a person in your address book. While a typical contacts app might include the person’s birthday, on the same timeline you could also include the publication date of the blog posts they’ve written. You almost certainly wouldn’t want such data included directly in the contact app, but surely such things should be discoverable from there.
When we talk about context, in a sense we’re talking about the given piece of information’s environment in the universe of data, a multi-dimensional ‘cube’. That cube could be seen to have different axes, like ‘people’ or ‘date’. So the journalist producing the news article will have colleagues and peers with congruent and opposing views (another axis) and a timeline of their articles, parallel to the timeline of events related to the story. Again, for practical reasons such information probably can’t be stored physically close to the story information itself. But it needn’t be, the Web provides a great infrastructure for a universe of data.
Chris asks for A comprehensive and CURATED compilation of related coverage and conversation in other places, but I reckon I’d argue that a fault of current news sources is that any curation is overly localised to the publisher. The alternative would be to have arbitrary curation, that could appear anywhere. Typically we’d want any filtering to be done to enhance relevance and trustworthyness within our own personal frame of reference. We’re more likely to trust friends and relatives than total strangers. This isn’t far from how Facebook decides what should appear in our Top Stories (compared to the more temporal Recent Stories). I find Twitter quite useful curation of links I might be interested in because it features the people axis (the social side) plus a timeline (which offers either recentness or novelty).
Another significant aspect is that much of the discovery/gathering of related coverage and conversation can be automated. Yesterday I found a good example of this in action. Check InsightNG, a very user-friendly online concept mapping tool (commercial but currently free in beta). When you add, say, a place to a concept map, it goes and looks that up in places like Wikipedia, and if you consider its findings relevant you can associate them with that item in your concept map (hence doing personal curation over the material).
I’ll take a moment to blow my own trumpet here, I’ve worked in this general domain for quite a while now on various different projects. My first major attempt was a desktop application I called IdeaGraph, a Personal Knowledge Manager. The core aim of this was very close to that on InsightNG, but I never got it ‘finished’, mostly due to geeky scope creep (did it really need an integrated Prolog interpreter?). A part of that problem was the end user ability to potentially create any kind of concept, and that proved tricky to manage.
As well as the underlying knowledge representation problem (which IMHO is well served by the (Semantic) Web), the user interface side is a biggy. Presenting something that is both user friendly yet can potentially support more complex operations (given user motivation) isn’t easy. Sweet spots are elusive.
Where XOW was intended for quick dirty input, IdeaGraph was intended to include a load of (new) information discovery mechanisms. Somehow I ended up back on that particular track this past year, working on NewsMonitor, essentially a feed aggregator with discovery & lots of linkability designed to plug into Fusepool/Stanbol, hence providing a source of related material for any app built on either of those projects. (Although the aggregator works, I haven’t quite finished the osgi integration bits yet, is high on my todo list).
[My Seki project is the core of my personal ongoing efforts in this general direction].
Anyhow, returning to Chris’ points, it is important to know the context of information, and the news angle is very significant given how understanding changes with new information (a more appropriate model of what’s happening can be discovered – something that’s arguably more likely in science than in politics…).
I’d better get on with something else right now (heh, one thing I discovered with IdeaGraph was Getting Things Done, ie. task management prioritization is my own key must-have for an information app). But I’ll leave some todo bullet points:
- raw vs. filtered
- individual as editor
- info overload
On the last point, remember Connolly’s Bane :
The bane of my existence is doing things that I know the computer could do for me.