07 Mar 2011
Watson, E-mail, and Semantic Technology
If you’re like most folks in the corporate world, you continue to be flooded with e-mails. Do any of these questions sound familiar?
- Can you e-mail that to me?
- Did you send that e-mail?
- Why didn’t I receive that e-mail yet?
- Should I email you that again?
- That seems to have bounced. Is the file too big for your email?
It’s no understatement to say that most large organizations still rely upon e-mail as their killer app. This holds true despite the fact that wikis, file sharing sites, microblogging tools, and their ilk are often far better communication mechanisms and repositories of information. In some companies, they have replaced intranets and knowledge bases anymore. They seem so, er, 1990s.
So, we know that e-mail is an important tool, even though it should not be the exclusive one. Weaning many people and companies off of e-mail is no small endeavor. As I have written about before, old habits die hard.
Improving E-mail
So, what can we do to improve things? How can knowledge be gleaned from millions of e-mails, rife with important–yet unstructured–data?
Technology seems to have created a monster but, as is often the case, technology can also solve the very problems it has created. In this case, semantic technologies are particularly well-suited for just this type of thing. We’re not just talking about simple, text-based searches for keywords–e.g., choice profanities or individuals’ names. We’re talking about much, much more.
Consider a recent New York Times’ article that sheds light on the increasing ability of new programs and technologies to interpret meaning from e-mails. From the piece:
…Cataphora software can also recognize the sentiment in an e-mail message — whether a person is positive or negative, or what the company calls “loud talking” — unusual emphasis that might give hints that a document is about a stressful situation. The software can also detect subtle changes in the style of an e-mail communication.
A shift in an author’s e-mail style, from breezy to unusually formal, can raise a red flag about illegal activity.
“You tend to split a lot fewer infinitives when you think the F.B.I. might be reading your mail,” said Steve Roberts, Cataphora’s chief technology officer.
Examples and Ramifications
The implications of the effective use of this technology are vast. Let’s say that you work in HR and have reservations about key employees defecting to one of your competitors. What if you could proactively discover and address their concerns, ultimately keeping them with your company?
Or consider a company attempting to launch a new product key to its future. What if you could determine the state of the product beyond formal status reports and project plans? What if employee verb choices could help you spot red flags?
Legally, at least many US courts have ruled that employees have an expectation of privacy while on the phone. However, they don’t have that same expectation while using company e-mail.
Simon Says
Stay tuned here. It’s going to be a wild ride. If Watson can beat supersmart humans on Jeopardy!, imagine what computers will be able to do with e-mails. I can think of good, bad, and ugly uses of these new technologies.
Feedback
What say you?
March 11th, 2011 at 1:43 am
Here they come again!
Software that “understands” but even more has the ability to “sense”! Watson and artificial intelligence. The claims that Facebook and Twitter were critical to the recent wave of uprisings! Now the proclamation that blogs and other forms of novel communications are better than e-mail.
The conclusion? Systems are telepathic! Watson with ESP!
Humans on Jeopardy are “supersmart” and by extension Watson is super-super smart? Give me a break! Watson is a series of algorithms according to its designers. Algorithms optimized for playing Jeopardy. Amusing, entertaining but please, let’s not go off the deep end and embody this mass of silicon with super human powers. The economy runs on algorithms and look where that got us! Financial chaos that no one can explain.
Algorithms gone wild. HAL!
And “semantic technologies? What are they? Essentially a Google search algorithm! Yes we are all being fed information based on an algorithm deigned and manipulated by Google. Others have been trying to “game” Google’s algorithm with some success so what does that tell us about the future of searches. We are being fed what Google wants us to read! This is beyond 1984. China wants to block Goggle and even our politicians are concerned. No more mind control! Ok, maybe I’m going off the deep end here but see how easy it is to embody all sorts of “human” qualities into technology.
And reading e-mails to get the sentiments of the senders and receivers? Sorry but that’s old news. More than 10 years ago I was doing work for a large global bank and the senior managers were concerned that when projects were being reported they were always on time, on budget but then suddenly a project would fall off the rails. Some searching on the history of the project exposed that fact that the project was going off the rails well before the crises. So I was asked how could the senior managers be forewarned projects that were in trouble. I suggested we examine the e-mails related to projects. Of course these “readings” were unbeknownst to the people on the projects. We would report on the “sentiment” as well as issues being discussed in the e-mails and report these directly to senior management. In the project review meetings the senior managers were aware of not only the sentiment but the issues relating to each project. We didn’t need fancy software to do this, just a few humans!
Software that attempts to assess sentiment has been around for decades. Some tried to build trading algorithms around this idea. Of course the problem is do you trade based on the “wisdom of crowds” or “irrational exuberance” or “the bubble effect”?
Imagine depending on software to assess the concerns of you employees based on their blogs and e-mails. Why not just ask them? Novel I know, but it doesn’t take software to determine whether your employees or customers are satisfied. Just ask them! They’ll tell you.
March 11th, 2011 at 5:31 am
Richord
Thanks for the comments. Interesting points.
Software that reads emails may be ‘old news’, but not everyone worked on those projects. Plus, there’s no way that that technology hasn’t progressed. I’m a big believer in asking people for information, but the entire point of sentiment analysis (as I understand it) is to glean what cannot easily be gleaned from unstructured data.
June 13th, 2011 at 9:53 pm
[...] be sure, semantic technologies will help us get our arms around enormous data sets, a subject that I have written about on this site before. This is already happening to varying extents. But technology alone won’t get us into the end [...]