Tuesday Belgium

Because he himself is trespassing in the exformation webspace, Lunghu occasionally checks in on Stijn Debrouwere’s blog to see what’s brewing (cross-lingual pun intentional) in the information architecture domain.  Stijn is a programmer/web designer/self-coined information architect based in Ghent, Belgium who blogs in both English and Dutch about the need to improve the ways that news content providers  on the web… um … provide news content.   Stijn posts even less frequently than Lunghu, which gives him time to actually craft some high-concept material and hammer home the relevant rhetorical points.

[Lunghu takes a different approach: for certain strategic reasons he usually prefers the scattershot, hit-and-run, all-over-the-map, death-by-a-thousand-cuts approach.  But enough of these trivial matters —read Sun-tzu if you would know why.]

Since Stijn’s posts are infrequent, Lunghu checks infrequently (RSS?  Probably not).   In mid-April Stijn had something to say about the importance (and difficulty) of providing context along with news content.   Some of his points intersect with aspects of intell analysis that are poorly understood within the intell domain and are even more poorly implemented just about everywhere.   Therefore, as a public service, Lunghu will first provide a gloss of Stijn’s views on news context and then will comment on the topic from the perspective of ongoing research into factors that contribute to information quality.   If boredom doesn’t set in beforehand, there may be time to outline the similarities with analogous problems in the intell analysis domain.   But maybe not.

Stijn’s main (only?) point in his April post is that news providers don’t do a good job of furnishing their customers with context for the news stories that stream past in a linear, no longer cyclical, flow.   Here’s some of what he says about information presented as news:  paragraph numbers from the original are included so that readers who wish to do so can check the verisimilitude of Lunghu’s characterizations.  [Material within square brackets is Lunghu’s addition, clarification or gloss.]

¶ 19:     [any news story is] “a tiny fragment of a long and complex issue.
¶ 23:     [journalists should be] “ … meeting people’s information needs.

But wait!  If journalists should be providing context in order to meet information needs, let’s define terms.   What does Stijn mean by “context?”   Since he doesn’t provide a clearcut semantic statement such as “context is X” we’re obliged to infer his meaning by reference to several descriptions of context that occur throughout his post.

¶ 4: [the information needed (context) is] “the chunky bits … the longstanding issues.
¶ 1: [Context corresponds to] “the background information [about a topic already] in [the readers’] brains.
[Lunghu notes that this sounds like a lot like “exformation.”]
¶ 16:     [Implicitly, context is the bigger picture]: “[some journalists] are telling the bigger picture through databases and visualizations.
¶ 13:     [Context is not merely] “additional background information.
¶ 11:     Context [is related to] analysis.
¶ 14:     (Elise Hu asserts that) [Context is] “the minimum you need to understand a topic … [plus more] as your need for more information increases.
¶ 15:     [Context is] “not just [what] information you provide, but also how you provide it.

These would seem to be rather widely dispersed attributes of context rather than a succinct definition.   Since we don’t have ready recourse to the seven Indian blind men who clustered around that elephant …  what IS this thing called context?

If Wikipedia doesn’t have the answer we want, perhaps academic researchers do.   Scholars at MIT and Univ Arkansas-Little Rock have been examining the topic of information quality for more than a decade:  what is it that makes information more —or less— valuable?   This is a question that’s more than academic:  it is (or should be) at the core of every web entrepreneur’s business plan.   But how many web designers have even heard of Wang, Strong, and Lee  –let alone know what they’ve discovered about information quality?  [Rhetorical question only.]  Read their papers if you would know … and/or read Lunghu’s gloss immediately below.


Lunghu’s gloss

Wang et al. have combed through two decades of research literature to identify (several dozen) factors that contribute to information quality.  These factors can be grouped into four categories:  intrinsic, representational, access, and –wait for it– contextual.   That’s right, context is just one fundamental dimension of information quality, and is itself composed of several factors or attributes.  Ignoring (for now) the representational and access dimensions of information quality, let’s take a closer look at the factors that comprise information’s intrinsic and contextual dimensions.  Intrinsic factors are those that inhere in the information itself, regardless of who uses the information or for what purpose.  Contextual factors are those that are contingent on who is using the information and how they’re using it.  Although various interested parties have proposed more than two dozen factors within these two information quality categories alone, Lunghu has whittled that list down to a manageable “top ten” set of information attributes (five factors each in the intrinsic and contextual dimensions) along with a brief description/definition.  Bear in mind that these are not mere binary factors that are absent/present, but instead each constitutes a scalar continuum from non-existent all the way up to optimal.

Factor Category Definition
Accuracy Intrinsic degree of conformance with actuality
Precision Intrinsic extent to which the parameters being described/ measured are defined/articulated with maximal specificity
Consistency{Coherence} Intrinsic degree of conformance w. prior information on the topic … OR … extent to which internal contradictions are absent
Credibility{Probability/Plausibility} Intrinsic extent to which the information can be believed
Objectivity Intrinsic degree to which reporting bias is absent > minimized >
explicitly defined.  
2 kinds: of the info source/
of the reporter
Note: specific numeric values assigned to the following contextual dimensions of information quality will vary depending on the needs of particular information users.
Currency Contextual degree to which information records an up-to-the-moment “snapshot” state of actuality
Timeliness Contextual extent to which information is available to the user when its content is needed [for decision-making or other processes]
Relevance Contextual extent to which information addresses the user’s implicit or explicit information needs
Quantity Contextual degree to which information contains all or most of the data elements needed by the user for the purposes at hand
Level of detail Contextual extent to which the information’s Precision matches the “granularity” needed by the user

This table of definitions doesn’t adequately convey the interactions between these factors, so Lunghu has helpfully provided a pair of diagrams that outline his view of the relationships that characterize the way intrinsic and contextual IQ are formed.   Of the intrinsic factors, and their cross-dimension interactions with contextual factors, we will speak no more … at least not right now.   For the moment, let’s focus on the contextual factors alone.  Note that Lunghu gives “relevance” the preeminent place among contextual factors:  information is relevant because it’s timely/current/at the proper level of detail and in the quantity that the consumer requires.   In other words, relevance is a supra-factor … in context.

However, what this means is that relevance of any information (news) item will vary from user to user/consumer to consumer.  Not every reader needs the same quantity of information or level of detail.  These will vary in inverse proportion to the exformation that each user already possesses.  Put another way, prior consumption/ retention of information on a particular topic (at various levels of quantity and detail) is what determines a big chunk of the consumer’s information needs … and her need (or not) for context.

[The nature of the decision against which the information will/may be applied is another major aspect/determinant of the user’s information needs.]

Quo Vadis?

Obviously, this view of information context has some rather significant implications for news providers  –or information providers of any kind.  If your (potential) users have wildly varying information needs, you’ve got unpalatable choices to make:  load up on background material for your lowest-common-denominator reader and thus alienate sophisticates; assume high levels of exformation and talk over the heads of the masses [the usual Lunghu approach], or attempt to steer a middle course and satisfy no one.

The good news is that there may be a way out –if content management technology can be imaginatively deployed in novel ways … and if information consumers are willing to specify their information needs to news providers in advance.  Think of it as “feed forward” rather than “feedback.”   Recall that contextual information quality factors can be thought of as “a scalar continuum from non-existent all the way up to optimal.”   The inverse/isomer of this idea is that customers’ information needs can also be situated along a scalar continuum  –one with multiple dimensions that roughly correspond to the attributes of intrinsic and contextual information quality.

Consider the graphic equalizer tool embedded/hidden in the interface of every installation of Windows® Media Player (granted, very few actually use it): ten slider bars adjust audio frequency bands from high to low, thus subtly(?) altering the user’s listening experience.  A similar “info equalizer” tool can be imagined that would allow information consumers to preemptively specify what levels of detail, quantity, currency, precision, objectivity, etc. they want in the content they seek.  These settings would probably vary from topic to topic for any given user, in addition to varying from user to user.  Default settings of the info equalizer tool might continue to deliver the mediocre news content we see today.  Some users (accustomed to Fox News) might want little more than maximal currency:  who cares about accuracy and objectivity?  Others might want to crank up the gain on level of detail and see what they get.  You probably get the idea already… no need for further examples.

On the content provider end, news items would have to be re-conceived as something other than self-contained information units existing in time and space.  Instead, it might be more useful to view information/news as custom-built niche products assembled on-the-fly from a multi-source, just-in-time logistics supply chain.  If each of the numerous sub-components carries its respective information quality metatags (visualized in the form of a radar plot?) from the cradle to publication, it may be feasible to match information inventory with customer-specified information needs.  Not easy, but feasible.

Before anyone starts sputtering “but-but-but, how do we do this?” let’s remember that Lunghu is a high-concept guy and is not about to start wrangling with all the niggling little implementation details that actually doing this would inevitably entail.  (Not unless his palm is crossed with silver and he hears the cash register ka-ka-ching!)  First let’s argue over/acknowledge the possible merits of this approach and then deal with the best way to make it work.  After all, we gotta slither before we can crawl.

Be seeing you!

Next time / some other time:  the implications for intelligence analysis.

Tags: , , , ,

%d bloggers like this: