E-Views and Reviews: Scopus vs. Web of Science

Alexander Street Press has acquired Classical Music Library (reviewed LJ 1/04, p. 176), and the possibilities for integrating audio with Alexander Street's text- and image-based databases are intriguing. The company seeks suggestions from librarians for new products combining various media. Respond to Eileen Lawrence's invitation at www.alexanderstreetpress.com/classicalltr.htm and tell them what you would like in a multimedia file. Meanwhile, I'm interested in your experiences with Classical Music Library and Naxos Music Library: Have you chosen one over the other, and if so, why?

Scholarly Googling?: If you haven't done so, take a look at the beta version of Google Scholar (scholar.google.com). Google is experimenting with this search system "specifically for scholarly literature, including peer-reviewed papers, theses, books, preprints, abstracts, and technical reports from all broad areas of research." The system FAQs highlight key questions, and we need to be assertive in making the answers known to our users (e.g., "My university subscribes to the Journal of Prosimian Dialectical Reasoning. How do I read the full text of its articles?").

Prediction for publishers: Give us something we don't get anywhere else, with simple, understandable search capabilities, powerful behind-the-scenes technology, and metadata that brings better and more relevant results. Deliver it all, and there will be great revenue for you in 2005. Seem like a no-brainer? Let's see what comes to market.

Quote of the Week: "Librarians realize…that manually intensive labor takes them away from the crucial intellectual work of creating collections and assisting clients in their use. So e-resource publishers are developing tools that offer more than content alone in an effort to minimize manual library work.

"The Books in Print and Ulrich's Periodicals Directory databases have been standard tools in libraries for decades, but recently we've been building enhanced tools around them. These new products enable librarians to better analyze collections against a core list of books and periodicals, as well as against the larger universe of publishing…. Tools like this are…freeing librarians up to do what they do best: think, lead, and direct.."
—Angela D'Agostino, Vice President for Business Development & Marketing, R.R. Bowker, New Providence, NJ

Scopus
Elsevier B.V.
www.info.scopus.com

What are they? Web of Science indexes over 8700 research journals across the arts, humanities, sciences, and social sciences, including over 230 open access journals. Scopus indexes over 14,000 scholarly journals in the scientific, medical, technical, and social science literature, including over 400 open access journals.

Web of Science consists of citations from Arts & Humanities Citation Index (1975–present), Current Chemical Reactions (1986–present), Index Chemicus (1993–present), Science Citation Index (1945–present), and Social Sciences Citation Index (1956–present). Scopus goes back as far as 1966, selectively. Both systems allow researchers to follow the development of research through cited searching (more on this later).

How Do They Work? The Web of Science search interface has never been my favorite, and even though Version 3's screen is an improvement, it's still busy and crowded. It uses terminology like, "Clear all search forms" and "Search using complex queries including field tags and set combinations," language that seems impenetrable for garden variety researchers. I wish there were a more direct route from the opening screen requirement that you select a General Search, Cited Reference Search, or Advanced Search to the next level of searching in each category. While what those first three options mean isn't clear, once you choose one, Web of Science makes it very simple to proceed, with razor sharp examples. ISI does use the term restrict for limit, but I'm quibbling, since limiting by language and document type is easy.

The Scopus search interface draws one in at first glance. It's easier to read (colors and fonts provide good contrast; it's less cluttered), it is oriented more toward topic searching than Web of Science, and the language uses less library-/ technospeak. The default Basic Search looks for terms in Article Title, Abstract, and Keywords, and the example effectively displays both nested and Boolean searching in a single phrase—a neat trick. It's immediately clear that this file is science-inclined—the default date range limit is from 1995 to the present. Plus, there is a limit to material added within the last seven days. This is available in Web of Science, too, but there you select your time span at the starting page at the same time you are choosing the type of search. I'd put the year limits at the point of doing the search.

Once you perform the search, however, all heck breaks loose on the Scopus screens. I have never seen one busier—and I remember the early days of Yahoo! You get four tabs for results— Scopus, Web, Patents, and Combined Results—which offer immensely useful ways of immediately refining your search: by source title, author name, year, document type, and subject area. The explicit Patent section will likely appeal to science and technology researchers.

The Web of Science search is a little less startling, with results looking more familiar. You can do the same refining of results: there's a quick "sort by" feature and an Analyze Results option that organizes the hits by author, source, and the other refinements listed above. You can search for Patents in Web of Science in a Cited Reference Lookup, but the system doesn't automatically isolate the results.

In Web of Science, you can easily do a Cited Reference Search for an author or a work, and your results are listed all together. They are not easy to interpret, however, since Web of Science uses the briefest of abbreviations to identify cited works. In Scopus, you search for the author and then get the cited references… attached to each article. There's a trade-off here: get an author's cites all in one shot (Web of Science) or get them article by article (Scopus). For the typical cited reference search, Web of Science delivers the information faster, but the Scopus format is easier for the novice (and most of us) to interpret.

Can You and Your Patrons Use It? My first search in both files was for "stem cell research." The Web of Science General Search found 478 relevant articles, the most recent of which was a November 2004 article in Nature Cell Biology and the earliest a December 1987 article in Acta Haematologica Japonica. My Scopus Basic Search found 670 Scopus results; 40,953 web results; and one patent. The most recent Scopus article was from a December 2004 issue of Journal of Surgical Research; the earliest was a December 1978 article in The Japanese Journal of Clinical Hematology.

I searched a series of topics ranging over medicine, engineering, chemistry, physics, and economics to compare the files: "HIV protease inhibitors" found 1,055 results in Web of Science and 5,568 articles in Scopus; "volumetric turbine flow meters" found no results in Web of Science and 15 articles in Scopus; "rotational angular momenta" found 18 hits in Web of Science and 958 in Scopus; "spontaneously broken gauge field theories" found no articles in Web of Science and 88 in Scopus; "economic globalization" found 202 articles in Web of Science and 2,725 in Scopus.

My searches for Seamus Heaney, ethnomusicology, postfeminist theory, and nihilist film all produced results in Web of Science but none in Scopus. No surprise since Web of Science includes Arts and Humanities Index and Scopus covers none of the humanities or arts.

Since author searching is a key element for these files, I did some extensive experimentation. I tried the usual vanity search and got unusual results. Web of Science pulled up 706 results on an author search for laguardia, c* (it indexes LJ). Scopus gave me more options for searching, and I received variant results: a Basic Search for "cheryl laguardia" in author names found nothing, but an Advanced Author Search found three articles from 1993 to 1994; a Quick Search for "cheryl laguardia" found one LJ article from 2002; and a Quick Search for "laguardia" found 58 articles—eight were by me, four of which didn't show up in any of the other three searches (but were from LJ). Also, the 2002 article had two different records. Caveat emptor when author searching in Scopus.

So I tried author searching for Nobel prize–winning chemist Dudley Herschbach. A Web of Science author search for "herschbach d*" found 298 results, all of which appeared to be relevant. A Scopus author search for "herschbach, dudley" found no documents; a Quick Search for "herschbach, Dudley" found one result in journals (and 1,706 web hits!); but the Advanced Author Name search for "herschbach, d" found 22 hits (along with 23 hits for "hershbach, d r").

To check results for social science authors I searched for John Kenneth Galbraith. Web of Science found 254 results on a search for "galbraith, jk*" while Scopus found 26 documents on an author search of "galbraith j k."

How Good Is It? After delving into these files I see that they are not trying to do exactly the same thing, so my ratings require context.

Scopus is a powerhouse for the sciences and seems to do a creditable job selectively in the social sciences. It's more suited for "subject" searching than Web of Science, and the content is more up-to-date. It includes web references but doesn't "reach back" as far as Web of Science (1966 for Scopus and 1945 for Web of Science), although for the sciences that may not be greatly important. Scopus rates a 9.75, with just one-quarter point off for the sensory overloading results screens.

Web of Science is still the only game in town for large-scale analysis of arts and humanities references and the major player for the social sciences. (ISI plans to offer a web citation index in 2005 that should compare with the web offerings of Scopus.) With its unique citing coverage of arts and humanities, Web of Science rates a 9 (the functionality takes one point off).

What's the Cost? Thomson refused to provide pricing information other than to say it varied. Scopus pricing is based on an annual subscription fee with unlimited usage. Pricing varies according to the size of the institution, between $20,000 and $120,000 per year. There is customized pricing for very small or very large institutions as well as consortia.

The Bottom Line: If you serve a primarily scientific, technical, and/or engineering clientele, you must get Scopus: it's big and it's good. If your clientele are predominantly arts, humanities, and/or social science researchers, you need Web of Science: it covers the scholarly material they need. But the reality is that many of us serve all these constituencies, and the hard truth is that we're going to need both files to satisfy them all.