Mathemagenic

Song, D., Bruza, P. D., McArthur, R. M., & Mansfield, T. (2006). Enabling management oversight in corporate blog space. AAAI Spring 2006 Symposia on Computational Approaches to Analysing Weblogs.

Abstract. When a modern corporation empower its staff to use blogs to communicate with colleagues, partners, suppliers and customers, the role of management in exercising oversight and guidance over g public speech of staff becomes dramatically challenged. This paper describes a computational solution to the interpretation of human-readable blog publishing polity documents into semi-automatic disconformance checking of corporate blog entries. The disconformance interpretation is regarded as an abductive reasoning, which is operationalized by information flow computations. Using a socio-cognitively motivated representation of shared knowledge, and applying an appropriate information flow inference mechanism from a normative perspective, a mechanism to automatically detect potentially non-confirming blog entries is detailed. Candidate non-confirming non-conforming blog entries are flagged for a human to make a judgment on whether they should be published. Experiments on data from a public corporate blog demonstrate an encouraging performance of the proposed methodology.

This is one of those in my big "to blog" list (more papers from AAAI 2006 Symposia on Computational Approaches to Analyzing Weblogs). I find this paper interesting for two reasons. First, the methods it uses could be interesting for our own work on topic detection in weblogs (see Anjo's blog, e.g. here). This one I'll leave to the specialists (in our team I'm more of a rough quality check: I look at the results of a new method and say if it makes sense given my subjective knowledge behind the dataset :)

The second is more for my own work: it's an assumption behind the research that what employees of a company say in their weblogs should be controlled by the company (hence developing methods to do so). Although I agree with the authors that weblog posts that do not comply with corporate communication policies can pose significant risks, I'm not sure I would ever want to work for a company that would try to censor my weblog. It's not the act of censoring that I'm against, but the idea that employees could not be trusted enough to judge.

This corresponds well with the recent post on Enterprise 2.0 Insecurities by Andrew McAfee (discussing the problem from another angle):

people already know how to behave appropriately, and they're not going to be driven suddenly wild by the appearance of the new platforms

That said, I have to admit that the issue is not an easy one. In one of my drafts I talk about it in a section on business challenges that blogging brings:

Lack of control of company’s message to the external world. Each blogger turns into "self-appointed spokesperson", communicating with the rest of the world based on own interpretations of corporate policies, interests and risks. This could turn into accidental leaks of confidential information, disrupt "official" public relations or marketing campaigns or create unexpected incidents when business-related information misinterpreted or amplified by media.

So, what might go wrong?

• Corporate policies (and other information necessary for a good judgment) are not necessarily explicit, easy to interpret or communicated well to employees. Sometimes it just doesn't make sense to think of any possible situation and describe it in policies at right level of detail.
• Corporate policies and/or bloggers themselves do not take into account specific characteristics of blogging that make it different from other forms of communication (e.g. speed, visibility, persistence) that could make a weblog post on a topic potentially more "risky".
• Sometimes "bad effects" of a weblog post couldn't be predicted in advance.
• Bloggers do make mistakes.

What could be done?

• Making sure that those rules that exist are clear and communicated to bloggers.
• Reflect on potential risks (especially those specific to weblogs) and raise awareness about them. Make sure that stories of bloggers about undesired effects of their own blog posts are heard by others. Do not police, but help people to learn making better judgment.
• If you really want to employ preventive text analysis try to turn it into an educating experience. E.g. instead of sending "suspicious post" to a manager for an approval, return it to the blogger with something like "there is a chance that you might be violating company policy regarding information about our stocks - are you sure you want to post it?". People do use spell-checkers to avoid stupid mistakes, so if positioned properly this might work.
• Think of potential benefits and relax. As said in Russia, those who do not take risks do not get a chance to drink champagne :)

As you could imagine, given my research I can't avoid doing it. A couple of years ago life was easy: you could always say "hardly anything has been published" and get away with it. Now things are more complicated, there is a lot of stuff all over the place

So, some of the challenges I'm struggling with:

Separating stories about what and how actually works from how things might or should be done. At this moment coming up with all kinds of normative or speculative ideas about business uses of weblogs is not a problem. The problem is that there are only a few studies of what actually happens in practice.

Figuring out how far you can (scientifically) trust a certain story based on explicit indications on what data has been collected and why.

Deciding how far I should go "out of the scientific" in selecting what to cite. Academic publications on business blogs are scarce, while there are quite a lot of white papers, case-studies from commercial companies, business publications or general media stories on the topic. And, of course, there are lots of ideas worth citing across the blogosphere.

The last one is a difficult decision. For an academic getting into research on business blogging it wouldn't be an issue: just run search through databases of scientific publications, work with the results and pretend that the rest doesn't exist. For me, learning about interesting issues in the field from weblogs years before something along the same lines gets "properly" published, it is a challenge. I can not pretend that the body of knowledge in weblogs doesn't exist, but, bounded by academic conventions, I can't figure a good way to fit it into my publications.

Even more, even if I try to give an overview of what is there on the topic across weblogs, I can't do it according to academic standards that aim for completeness and objectivity. I know that I shouldn't even try to provide a complete and objective picture when giving an overview on post on whatever issue across weblogs.

Why it's so incredibly easier and more rewarding to write a blog post than a piece of scientific text? Why can't I have best of both worlds: grounded claims of academic publications and the style of a blog post?