Reference Reviews

Péter's Digital Reference Shelf

September 2006

Title: Annual Reviews Archive
Publisher: Annual Reviews, Inc.
URL: http://annualreviews.org
Cost: see details in the review
Tested: August 2-17, 2006

The Context

When more than 70 years ago James Murray Luck, a biochemist and a philanthropist, published the very first issue of the Annual Review of Biochemistry, he may not have thought that he would establish a special genre, and an extremely successful one. I reviewed the AR archive four years ago, but there are new features both in content and software, as well as a recently announced sweet deal for back issues which more than justify its review again.

Since the first volume, Annual Reviews, Inc. has launched 30 more Annual Review titles in the series, and the only one which apparently failed was – interestingly – the Annual Review of Computer Science. Several disciplinary series of the Annual Review family have been the most respected in the discipline, and top-ranked ones year after year in the ISI Journal Citation Reports, in the respective categories.

In 2005 half of the Annul Review series were ranked no. 1 from Astronomy & Astrophysics to Plant Biology. Many others were among the top five, often in the second or third position, such as the Annual Review of Microbiology, Annual Review of Nutrition, Annual Review of Physiology. No wonder that the Annual Reviews, Inc. Web site proudly publishes (and updates) the impact factor of its series.

The few Annual Review series in the social sciences (except for the Annual Review of Political Science) are also among the top-ranked serials by impact factor. The Annual Review of Psychology has been a perennial no. 1 among the 100 journals in the Multidisciplinary Psychology category, and no. 1 or no. 2 among all the nearly 500 titles of the broad psychology subcategories combined (Social Psychology, Applied Psychology, etc).

The rank order of an Annual Review series may depend on the category into which it is grouped. Several AR series are listed in multiple categories. For example, the Annual Review of Public Health is ranked no. 1 among 63 serials in the Public, Environmental & Occupational Health category within the social science domain, and no. 10 in the very same category of the science domain (which ranks 99 serials). Some are ranked no. 1 no mater what the domain and/or the category, such as the Annual Review of Pharmacology and Toxicology. Either way you look at it, these ranks tell a lot about the influence of these publications.

There are a few other Annual Review of ... publications by different publishers who were encouraged by the stunning success of the series, and started their own. Some have given up after a few years, such as Annual Review of Fish Diseases; others are doing fair, such as Annual Review of Nursing Research; or doing pretty well, such as Annual Review of Applied Linguistics; but none have the same clout as the ones published by Annual Reviews, Inc. -- with one exception. The Annual Review of Information Science and Technology has been among the top-ranked serials by impact factor in the Information and Library Science discipline for decades, often #1, sometimes #2 or #3.

The Content

The Annual Reviews series are among the best ready-reference sources not only because of their outstanding, evaluative, succinct content (written by the most knowledgeable experts in the particular subject field, who as chapter authors in turn get additional clout), but also because of the immense value of the hundreds of references at the end of each chapter, citing the most important articles, books, book chapters, conference papers and government documents in the subject area published in the previous few years.

There are records for more than 20,000 chapters in the Annual Reviews archive. The bibliographic records and the abstracts are open access. However, abstracts are available only for chapters published after 1994.

It would be a good idea to include the lead paragraph from the full-text documents for the pre-1994 chapters. It would be informative to see the excerpts in scanning the result list, and it could also improve the precision of the searches. Theoretically, the abstract produced by a professional abstractor is better than the lead paragraph alone, but practically, the latter method is used gingerly by many of the indexing and abstracting services for a large number of records in their databases.

It is quite reasonable to assume that if the abstract includes the query term, the chapter is more likely to discuss that topic represented than the other chapters retrieved by searching the full text. These may mention in passing the query term. For example, searching for the the topic of prejudice, the term may appear as "without prejudice". Currently, the truncated query term prejudic* finds 10 hits when limited to the abstract and 171 hits when searching the full text. The latter is too many from this archive for being a central topics, while the former result set may miss some important chapters.

The number of records is modest compared to many other publishers' and digital facilitators' archives that offer open access to more than 100,000 bibliographic records. However, the majority of the Annual Reviews chapters quickly become core source documents for the topic, as proven by the number of citations many of the chapters receive. While there are a very few chapters that don't get cited, the average chapter is much more cited in most series than the average individual articles in journals of the discipline, including the best journals. That is clearly what the ISI impact factors indicate and why I ruminated over the impressive impact factor of most Annual Review series in the Context section.

One of the most valuable content features is that AR provides free information about the citedness score of the individual chapters in the ISI Web of Science (WoS) database which indicates the importance of the chapter. (HighWire Press, which was the previous digital facilitator of Annual Reviews, Inc., pioneered this gem of a service years ago and offers it for hundreds of the journals it hosts).

This service does not require that your library subscribe to the WoS database. The score is displayed along with the bibliographic data on the details page of each chapter. The language of the message "Citing papers via ISI Wb of Science (37 or more)," may suggest that the data is not updated instantly, but in my test I found them as current as WoS. But this is just a starter.

Clicking on the citedness score entry will display a list of citing papers, taking the user to a special service area of WoS. Only 20 citing articles are listed for non-subscribers of WoS, but even with this restriction this is an awesome piece of value added information. It is yet another bonus that the ISI master records for the max. 20 citing papers can be also displayed for free.

There are several other links, but I will discuss those in the software section. I decided to bring up here the WoS citedness scores because they are already part of the AR records' content when the records are displayed. If there are no citing papers (such as for prefaces, errata and very currently published chapters) then the option is not displayed at all, so the absence of that entry in the sidebar makes it clear that the chapter has not been cited (yet).

The list of cited references is part of the full-text documents available only for subscribers of the online edition. Looking up such a list is also a rewarding experience as many of the cited references have links that take you to at least the free bibliographic record of the cited paper through CrossRef, PubMed or ISI, or all of them. These then may provide abstracts and or links to an open access version of the cited paper. These, in turn, may offer their own extras, such as the citedness score of the cited paper itself -- received from papers in journals within the archive of the publisher or its digital facilitator, such as Ingenta Connect, which hosts the journal in which the first cited paper was published. Of course, the citedness scores will be different in WoS, CrossRef and Ingenta Connect as they are restricted to the domain of journals in the archive of the publishers' or of its digital facilitator.

The digital edition also has supplementary materials that could not make it into the print edition for various reasons, such as sheer size (such as long survey questionnaires, bibliographies of large meta-studies discussed in the chapter), or a video clip of chimpanzees (which were the subject of a chapter in the Annual Review of Anthropology). Meta-studies of surveys and experiments are becoming very popular in some disciplines (and on a different scale imitate what the AR series were set out to do for original research publications).

The Software

The new software looks very similar to the earlier one (by Highwire Press), but it is from a different company, Atypon Systems, Inc. Indeed, "imitation is the sincerest form of flattery" and in this case it is to be applauded, because HighWire Press has the best software for bringing out the most from publishers' archive.

The software provides many links from the master record page of the chapters. In case when there is no abstract for the record of the chapter, these links for extra services and content enhancement literally stand out even in the sidebar.

Beyond the free citedness scores of the chapter in WoS and other WoS-related information, the most important link is the one which shows all of the citing references from articles whose metadata was submitted to CrossRef. This is not only a very useful bonus but also a natural one, considering that Atypon Systems, Inc. designed the elegant and smart Web site of CrossRef. I wish Atypon had been involved in the gathering and processing of CrossRef partners' metadata rather than Google because the former recognizes so much better the citing-cited paper pairs than the latter. Google often takes you to a wild goose chase following purportedly citing articles which do not cite the document at hand. See my column in Online Information Review about the phantom citation counts dispensed by Google Scholar and my presentation at the closing plenary session of the UKSG conference. I could not find any phantom citation in the CrossRef citing items list in my tests. Laudably, some other publishers also have started to use the CrossRef database to enhance their services, by looking up citing papers in the CrossRef database.

In a sense the CrossRef citing items are also part of the AR master record as the most current ones are automatically displayed after the abstract of the chapter – you just don't see them unless there is no abstract. Atypon could make this even better by informing the user right in the sidebar, how many CrossRef items have cited the chapter at hand – just like it does for WoS. I understand that in order to display the entire list of CrossRef citing items may require an extra link, but as of now, the highest number of citing items was around 25, so overloading of the system by producing a long list of citing articles is not an issue yet.

There are additional link services to find the related records in CrossRef, PubMed Central and WoS based on common cited references among documents in the collection. The related items can be indeed very closely related to the chapter at hand, but usually only if the overlap between the set of cited references of the "relatives" is higher than 60%. The relatedness is well explained, the overlap is perfectly clear in WoS but not in the other systems. The software offers e-mail alerting service when new item cite the chapter, downloading the bibliographic data to a bibliography management software, e-mailing the record to another person, along with a few more common options.

Given these excellent digital content and software features it is pennywise and pound-foolish not have the digital version and the print versions of the volumes. The volumes are priced between $85-95 for the individuals, and about 2-2.5 times as much for institutional use. The print editions have the same price as the online editions and the big deal is that the combined print-online editions for institutional use cost only 15-20% more than the combined print-online editions. Individuals get both the print and online versions at the same price below $100. The latest news is that for a one-time price of $5,000 the Electronic Back Volume Collection can be acquired of all the volumes of all the series published before 2000. This may be the best investment of your library's shrinking serials budget.

— Péter Jacsó