On October 23, Eliyon Technologies (http://www.eliyon.com) launched Eliyon, a new product that uses Web data-extraction technology to build a database on the business history of 6.4 million people and 560,000 companies in the U.S. According to Eliyon public relations director Jennifer Lichtman, this product was under development for 2 years prior to its release. The company feels its main competitors are Hoover's and Dun & Bradstreet.
Eliyon Technologies was founded in 1999. Based in Cambridge, Massachusetts, it's affiliated with Corex Technologies Corp., a leader in business card scanning software and hardware. Jonathan Stern, Eliyon Technologies' president and CEO, holds the same positions at Corex.
Eliyon's system crawls the Internet 24 hours a day, reading news articles, press releases, company Web sites, and SEC filings. The entire database is updated every 2 to 3 weeks. Only free sites are crawled and sites with .edu, .gov, and .mil extensions are excluded. Theoretically, pornographic sites are also excluded by looking for matches to particular words. Using natural language processing tools, the system extracts information about people and the companies for which they work and combines the individuals found in multiple places into consolidated profiles. There are no human editors or human intervention; the entire process is automated. This, unfortunately, is the cause of some of the product's pitfalls.
When Eliyon's system finds information about a person, the page on which the information appears is cached and kept in the system forever. Therefore, even if the original page disappears or moves, the cached version always remains available through Eliyon. A notation is added at the top of the page that includes the date and time the page was cached and a link to the current version of that page, assuming it still exists. When asked about the copyright issues involved in the caching of pages, Lichtman felt there wasn't a problem, as the entire page is cached without modification and Eliyon inserts its notation at the top of the page.
Eliyon may be searched in the following ways:
My preference would be to default to the Power Search, as all options are available through it.
- Job Title
- Biography, which includes company name, description, and location; work history (current and/or prior companies); education institution attended; degree attained; and area of study
- Alumni Search, which finds individuals based on their educational background
- Name, which also includes title, company name, and work history
- Company Search
- Power Search, which incorporates all of the above criteria
As a professional searcher begins to use the system, some glaring weaknesses appear. The first problem is the lack of a controlled vocabulary for any of the searchable fields; all entries are free text. This begs the question of how to account for variations in title from one company to another. For example, if you want to find the names of the chief financial officers (CFO) in a particular industry, that would be fine if chief financial officer were a standardized title. While CFO is commonly used, so are controller, comptroller, vice president of finance, and so on. How would you capture all of those people using Eliyon's system?
As another real-life example, I searched for my own name. Fifty-seven matches turned up, although many of them were duplicates. (There's also no way to tell if an entry is a duplicate without viewing each individual entry.) Of those 57 entries, one title showed up in 40 of my entries: "Experienced International Business Researcher." While I appreciate the recognition, I would hardly consider that to be my title. All of these entries were drawn from the following reference to my book: "written by experienced international business researcher Sheri R. Lanza and edited by Searcher magazine's Barbara Quint."
Another entry had my title as "Treasurer, Budget and Finance Committee Chair, Membership Committee Chair." This apparently was a reference to the fact that I serve as secretary for the AIIP's board of directors and the words "treasurer," "budget and finance committee chair," and "membership committee chair" were in close proximity to my name. The rest of my 57 entries didn't include a title. So the fact that my "real" title is president of my company, Global InfoResources, Inc., never even made it into the Eliyon database. I also never showed up as a writer, columnist, speaker, or consultant, all of which could be considered valid titles.
A similar, and potentially more serious problem, is the lack of a controlled vocabulary for industry. Per Eliyon's Web site, "We allow you to match the same keywords used by a company to describe their products & services in press releases and on their website." Unfortunately, this makes the assumption that all companies within an industry will use the same words. While that's possible, it could be quite dangerous to rely on that supposition.
In fairness to Eliyon, the database wasn't designed to conduct a search on a person's name. As Lichtman rightly told me, if you already have a name, it certainly makes more sense to use a search engine to find information on that person. Instead, Eliyon was designed primarily to look for names to attach to a particular title. In that case, Eliyon would be more suitable for executive recruiters, human resources personnel, sales/marketing professionals, and development organizations looking for potential fundraising contacts.
I asked Lichtman what the fee-based Eliyon service offered that the major free search engines (Google, AllTheWeb, AltaVista, etc.) didn't. She replied that Eliyon has a value-added benefit in that a search returns "specifics" about a person. I discussed this further with fellow researcher Gary Price and we both came to the same conclusion: An experienced searcher, using a good search engine and a focused search strategy, would likely receive results as good as, if not better than, those returned by Eliyon.
Currently, Eliyon is priced at $99 per month for unlimited usage. No contract is required so one could sign up to use it just for a month while working on a particular project; there's no long-term commitment. However, this is introductory pricing and it's possible that the fee will increase when the introductory phase ends.
While I'm aware that I've pointed out many problems with this system, with some changes and tweaking it could have potential for the information professional. Adding controlled vocabulary for titles and industries, removing duplicates, adding human intervention at some point in the process to help to eliminate "nonsense" entries, integrating a protocol for handling name variations, and addressing the potential copyright problem of caching Web pages would be a good start. With those changes, Eliyon might be a useful product for information professionals.