InfoInnovation

Scientific/engineering and cultural content moving toward the Semantic Web

Monday morning at the NFAIS conference: the first session presented three fascinating examples of next-generation content enhancement approaches. Call them Semantic Web, Web 3.0 or what have you, each offered signposts for broader trends in the coming information content environment.

Most striking (and commercially significant) was today’s announcement (and the presentation by Rafael Sidi, vice president of product development for Elsevier's Engineering and
Technology Division/Corporate Markets) of illumin8, the result of a partnership between Elsevier and NetBase, Inc. From the press release:

illumin8 combines search and semantic indexing technologies to distill deep meaning, purpose and insight from…Elsevier’s full-text content, scientific abstracts from 4,000 publishers, patents and billions of web pages. This research tool extracts and analyzes solutions, which are then categorized under organizations, products, technologies, approaches, and experts. illumin8 is designed to go beyond simple keyword search, quickly finding and extracting crisp summarized answers and interrelationships that are semantically related to the context of the search query.

Krista Mantsch, senior research librarian at the National Geographic Society, provided an impressive overview of the many ways in which her organization is applying Web 2.0 technologies to the extension of the library’s services across the Society and its partners.  Most interesting to this blogger was the one externally-focused project she discussed, called Metalens. The result of another technology partnership, with Clear Path Labs, it is a beta initiative that uses semantic techniques to tie media and metadata to maps - connecting the “where” to the “what” and the “when,” to create a "bigger picture" (such as the travel narratives that NGS staff are beginning to create) from small, related media assets.

Finally, Martin Kalfatovic, head of the new media office and preservation services at the Smithsonian Institution Libraries, described the Biodiversity Heritage Library (BHL) project, created to support the Encyclopedia of Life inspired by E.O. Wilson and launched last May. The BHL is a collaboration among several major research libraries and museums involved with biology and botany. It is developing a corpus of some 300 million pages of relevant literature being scanned with the help of the Internet Archive. Of particular interest is a focus on creating “taxonomic intelligence,” using automated structural markup, semantic markup with geographic and taxonomic tags, and a variety of persistent identifiers. The BHL is seeking to work with publishers of copyrighted content, offering the advantages of scanning, structuring and managing their content, exposing it more widely, and imbedding it in related knowledge. To date, though, only a few small publishers are participating.

GlobalSpec: detailing the "what" and "how" of a vertical search role model

In any discussion of vertical search, GlobalSpec is bound to come up as one of the prime role models. Though it is privately held, and funded primarily by Warburg Pincus, its rapid growth has been well-known (and widely envied) for  years. One of the highlights of Infocommerce 2007 was a detailed presentation by chairman and CEO Jeffrey Killeen, rich in both the statistics and the philosophy behind the company’s ongoing success.

According to Killeen, the story (like many Web successes) began with seeing the Mosaic browser - and recognizing the potential link to catalog data. From then on, there was a consistent tripartite focus – engineering, workflow, and e-media. The initial product, SpecSearch (parametric search on electronics parts catalogs) is now the largest database of its type with more than 2 million products searchable. Two years ago (it seems longer – perhaps because of the high profile of vertical search in that time period) came the Engineering Web – a curated collection of websites, tied together by a proprietary taxonomy, that was originally created to drive more frequent use of SpecSearch (which, like all GlobalSpec offerings, is advertiser-supported and free to users). There are now 3.6 million registered subscribers, 3 to 4 million unique monthly visitors, and (from a beginning just two years ago) 5.9 million opt-in newsletter subscriptions representing 2.3 million unique subscribers. The result is a rich environment for hard sales leads and contextual advertising.  GlobalSpec has further capitalized on its premier position by creating an ad network providing additional inventory to its advertisers. Killeen quoted a five year compounded revenue growth rate of 70%, positive cash flow for the last two years, and progress (“half-way”) toward a target EBITDA of 35-plus percent.

Underlying this success have been some consistent principles, chiseled in stone. “Things we are intensely commited to”: 1) a registration model optimizing the value of clickstream data, built with careful attention to the value exchange with users (registration in exchange for increased productivity), patience, and derivative businesses (e.g. e-newsletters); 2) a default to user-centricity in any trade-off between user and advertiser needs; 3) a “what are we solving for now” management approach that has followed multiple life cycle phases (product model/vertification; early audience build; cash flow break-even; audience/advertiser scale with sales force expansion and new product rollouts; and EBITDA increase via flat headcount and continued high degree of innovation); and 4) “managing from the perspective of a blank sheet of paper.” “Things we said no to”: software as a product, e.g. product lifecycle management (PLM) and product data management (PDM) a la IHS; user-paid models; price-cutting; private-labeling; other verticals (“for now”); “weak” M&A opportunities; “premature” strategic sale.  When that sale does come, it will be one of the great value-creation stories of the post-Mosaic information industry.