Elsevier and NetBase Launch illumin8
Posted On February 28, 2008
This week, Elsevier (www.elsevier.com) and its technology partner, NetBase Solutions, Inc. (www.netbase.com; formerly Accelovation, Inc.), have launched a new research solutions information service called illumin8 (www.illumin8.com). The new service uses text analytics and natural-language processing technologies developed by NetBase for its Tech Scout service, previously in use at several dozen Fortune 500 companies. The illumin8 service taps some 5 billion websites, blogs, and forums, 3 million full-text articles from 1,800 Elsevier journals, 33 million article abstracts for 15,000 journals from some 4,000 publishers, and 22 million patents from five worldwide patent offices. The new illumin8 differs from the late 2007 version of Accelovation’s Tech Scout in the application of a sophisticated end-user interface focused around solutions needed by knowledge workers in R&D and marketing research departments and by Elsevier’s direct marketing of the service. While Tech Scout will no longer be offered, NetBase will continue to produce independent enterprise services, e.g., a "needs"-oriented service called Accelovation for now, while also enhancing and marketing illumin8.
Elsevier is promoting illumin8 as a revolutionary productivity tool for research and technology intelligence. Rafael Sidi, vice president for product development at Elsevier’s engineering and technology division states, "R&D knowledge workers no longer need to spend hours, days, or weeks sifting and reading through thousands of search results or full-text documents, and then manually synthesize results to find actionable answers. illumin8 is poised to transform the way companies innovate, find solutions to their R&D problems, collaborate and make key R&D investment decisions."
Commenting on NetBase’s technology and its application in illumin8 and Tech Scout, Sue Feldman, vice president of search and digital marketplace technologies at IDC, described the process of text analytics. "It’s not search. Search asks questions and matches them to all the documents with that term or that concept. Then it takes all documents and relevance ranks them by a variety of parameters. What text analytics does is it merges all the documents into one big collection, but not as documents. It’s looking mostly at sentences and paragraphs to extract information and put it into a standard table. The framework extracts certain occurrences of types of information to fill the frame. NetBase looks for problems and solutions."
The illumin8 interface extracts "solutions" from its massive content flows and categorizes the abbreviated results into products, organizations, approaches, and experts. Searchers can reach content by Topic or broad solution areas, Problem, and Benefit. When they find content of interest, users can push through to the full-text references as citations and, in the case of the Elsevier publications, whatever they subscribe to through Elsevier Science Direct. According to Sidi, any user who doesn’t have access to full text through a Science Direct subscription can reach a pay-per-view option, e.g., on publisher websites. However, many users will probably stick with the extracts and then network with organizations and experts. Graphic displays can sharpen the impact of content on users. For example, organizations are displayed in bar charts in declining order of importance.
The profile view in illumin8 also shows the type of literature tapped in response to a query—journals, blogs, government, corporate, company websites, author names, etc. Then with a detailed view, users can go to the full article. Users can also refine search results by approach/product/organization/other, type of content (journal/patent/website), an accuracy versus quantity slide (aka precision versus recall), or by solution type.
No information tool is better than its content, as everyone—well, everyone reading this NewsBreak—knows. Nor is any tool designed to manipulate content better than its metadata. For example, clicking on an organization’s bar in the graphic display will take users to content from that organization. However, when I looked at one display in which the "Department of Energy" was by far the leading organization, I noticed that further down the graphic appeared the "U.S. Department of Energy," then the "US Department of Energy," and further still, "Argonne National Laboratory," a division of the DOE. Sidi told me that they already had plans in place for "normalizing" organizational names.
But what about the most difficult of all literatures—patents—with their obfuscatory language often designed more to conceal than reveal? Mary Poul, marketing manager at the medical division of 3M and a self-defined "champion" of the old Tech Scout as well as the new illumin8 service, indicated that at 3M they did not turn to illumin8 for patent searching. "We have full-time patent searchers at 3M. When we did the rollout for Tech Scout/illumin8 in November, we didn’t highlight the patent searching. We focused on solutions."
Founded in 2003, NetBase’s sophisticated semantic indexing technology, which was developed at MIT’s Media Lab, powers illumin8. NetBase also does its own web crawling for the service, supplemented with some content from Elsevier’s Scirus web search engine. The Tech Scout product from NetBase, which illumin8 replaces, included market-needs scouting, business models, and needs statements. According to Poul, the new illumin8 has a better interface, one that allows end users to use the service directly, instead of having to go through librarian searchers as with the old Tech Scout service. Poul did mention that the original version of Tech Scout also emphasized "need statements, not just solutions, what problems people have." She expected that service to return in future versions. However, according to Jonathan Spier, NetBase’s CEO and co-founder, the service to identify "unmet market needs and tap the Web for consumer insights" will fall into the Accelovation service marketed by NetBase to enterprises. According to Spier, NetBase plans to phase out Accelovation in time and introduce other services for enterprises.
How much does illumin8 cost? According to Sidi, the base price for a Fortune 1000 company starts at $100,000 a year and then builds based on the number of users (FTEs). I posed the question to Sidi that, while targeting Fortune 1000 companies made business sense for a new product, the product could have revolutionary impact on smaller, medium-sized companies that weren’t as information rich as giant corporations. Sidi responded, "It is our plan to version this product for that market. We don’t want the versioning to filter based on subject, but on features and functionality. We want to offer the option to use the full version."
Jens Tellefsen, vice president of marketing and product strategy at NetBase, added, "It won’t be hard to envision future capabilities for this product—visualization, interrelationships between technologies, some more premium features." Spier was excited about the visualization possibilities. "One of the great things about this type of analysis is that once you have the data categorized, you can build any kind of visualization." The visualization should find eager reception. When I asked Poul what she wished they would add, she replied, "Some neat macro level reporting, a topographical map of the search like Vivisimo clustering. It would make search results more attractive and draw people into more digging to make sense of the results. Maybe even mind mapping or theme mapping."
What does the future hold for illumin8? According to Sidi, they plan to introduce internal enterprise collaboration features in Q2 2008, including the ability to connect to outside experts and research institutions. He also indicated that the future might include the integration of citation information to measure the impact of experts, technologies, and organizations for a "360 view." Tellefsen says they plan to add additional features to illumin8, as well as develop products for other corporate user groups, e.g., marketing departments. Improvements will include the ability to merge and share results in different combinations, according to Tellefsen, as well as adding content on licensing and partnership opportunities. They already reach lots of open access sources, e.g., through deep crawling of around 1,000 top university sites. Spier indicated that they plan to introduce feedback features for illumin8 users that would include a blog (http://blog.illumin8.com), a user forum, and ratings for searches and individual data elements. "We already have these in Accelovation. They should appear in illumin8 later this quarter or Q2. We have a list of enhancements longer than my arm. Feedback is valuable."
As for the potential success of illumin8, Feldman states, "The very best information system has to have good content, what the users would expect for the topic, and good technology. The other half of the system is the educated user. The bootstrap goal here is to get from so much information that users can’t possibly handle it manually to giving people tools to understand what’s going on that they can use to analyze information in ways that would be impossible manually. All these technologies are trying to sift information so people can pay attention to what brains can do better than machines."