Pulling out the heat chart (FF pt. 2)

As noted in an earlier post, there were two big stories at FASTForward08 -- and I'd suggest there's a third positioned to play out. The first story is the technology - advances in search are claimed that have some of the players speaking of "visionary technology" (that's Zia Zaman of FAST); the second story is capital - how these advances will yield new modes of profitability and competitive advantage. The third not-quite-written story has to do with the relation of these two: of technological depth and business technique; visionary promise and practical payoff.

I'll try to offer in this and a few more posts a few glimpses into what's at work. It's fascinating stuff, it's also very complex. I'm neither an analyst nor do I pretend to have special competence in corporate fields such as Knowledge Management. For some sharp appreciations of the implications of what was on offer at FASTForward08 for KM, for Business Intelligence, social computing and the like, check out some of the conference bloggers who are writing about them here and on their own blogs. Here at IMproPRieTies, it's all amateur all the time.

Among the vendors I met at the conference were Brian Pinette (left) and Tim Mohler, of Lexalytics. Among its skills, the small Amherst, Mass. firm offers a very specific expertise: They search texts -- blogs, movie reviews, restaurant critiques, political discussions and much else -- with the aim of gauging sentiment.

Two things to note: They use a form of text analytics that derives from unstructured search. Structured search categorizes and indexes databases and objects that have clearly demarcated fields of pre-set sense. Unstructured search crawls across all sorts of things - texts, music, video - with the view of pulling out from the massive amount of stuff that whatever happens to be meaningful relative to the purposes and needs of the searcher.

According to their White Paper on Sentiment, Lexalytics first parses texts into parts of speech, then looks for emotional language, often adjectives, and examines the proximity of the emotive words to key terms. "Devastating loss" offers a pretty high negative, but the context will be examined for other similar language in order to compile a case for the strength and positive or negative valence of the sentiment.

Often texts address more than one subject -- a political blog might weigh in with a comparison of two or more candidates and their positions vis a vis one or several issues. Lexalytics (and similar firms: see this description [link no longer active] of customer voice analysis by Palo Alto-based Attensity) [see this update on Attensity[link deleted, no longer active]] says it can parse sentiment by sentence or other localized segment, or by entity. It can tell, for example, that blogger "abjectNEOCON" strongly prefers one candidate to another; it can sample hundreds of such blogs and offer an overview of the bunch (think of the marketing application); it arrives at numerical gauges of sentiment about products, companies, feeling threatened or safe, and the like), and locates them on a graph, or "heat chart":



The new thing here is not that one can go about assessing consumer sentiment. The new is that one no longer needs to do it "by hand," or by structured (and hence "artificial") question and answer, or poll-type data collection devices. The candid, naked sentiment out there in blogs, newsgroups, chats, Youtube, twitter -- the entire inchoate realm of information -- is now susceptible to highly sensitive and sophisticated modes of analysis.

I can't help being reminded that this eventuality was anticipated by Locke, Weinberger, Searls and Levine quite some time ago:

6. The Internet is enabling conversations among human beings that were simply not possible in the era of mass media.

7. Hyperlinks subvert hierarchy.

8. In both internetworked markets and among intranetworked employees, people are speaking to each other in a powerful new way.

9. These networked conversations are enabling powerful new forms of social organization and knowledge exchange to emerge.

10. As a result, markets are getting smarter, more informed, more organized. Participation in a networked market changes people fundamentally.

Let's not forget # 12: There are no secrets.

More later.