Péter's Digital Reference Shelf
September 2002

Title: DOE Energy Citations Database
URL: http://www.osti.gov/doeecd
Publisher: Department of Energy (DOE)
Cost: Free, registration may be required
Tested: August 20-28, 2002

Note: This review is about the DOE Energy Citations Database (DOE ECD) hosted at http://www.osti.gov/doeecd. There is a much smaller version of this database at http://www.osti.gov/energycitations with a confusingly similar name: Energy Citations Database (abbreviated ECD, without the DOE prefix). Even the two logos are very similar. The former database is meant for the DOE and contractor community and is supposed to require registration. As this is mentioned only in a background file, I was not aware of it and I could freely use the database. I also tested the smaller Energy Citations Database and have included information about the number of hits, in case you can not access the larger database.

I find it unfair that the DOE Energy Citations Database received far less attention than its much inferior elder sibling, the PubSCIENCE database, which I have reviewed separately. True, PubSCIENCE has been around for 2.5 years, while DOE ECD was released only 10 months ago. Seniority, however, should not matter in this case. After all, the core of both databases is derived from the Energy Science & Technology (ES&T) database that has been available on DIALOG for a long time at $75 per hour and $1.65 per record. Both databases are only subsets of ES&T, but few users will miss the materials that are available exclusively in the commercial version.

PubSCIENCE has the supposed advantage of providing links to publishers' archives, but as I demonstrate in my other review, that advantage is minimal for several reasons. One is that the links are often to the home page of the publisher, not to the article abstract. The other is that at the linked site you will either not find the abstract, or you find the exact same abstract that is in the PubSCIENCE or ECD/DOE ECD record. Read the companion review to learn about many of the other deficiencies of the PubSCIENCE database.

Content

The scope of the database is somewhat broader than you might expect from an energy database. It "includes bibliographic records of literature in disciplines of interest to DOE such as chemistry, physics, materials, environmental science, geology, engineering, mathematics, climatology, oceanography, computer science and related disciplines. It includes citations to report literature, conference papers, journal articles, books, dissertations, and patents." Also, one disciplinary area that is nicely covered, but not mentioned, is biomedicine.

However, when you search the databases, it denies having theses/dissertations in ECD and DOE ECD. Still, I did find some dissertations in the results of the 45 test, like this Dutch-language one.

As for the size of the database, ECD has a little more than 2-million records, while DOE ECD has close to 3.7-million items. Both dwarf PubSCIENCE, which offers only 1.3-million records. One of the reasons for this is that the majority of document types in PubSCIENCE is journal articles. But even with that restriction, DOE ECD is far ahead with 2.5-million records for journal articles alone.

This ratio is also well-reflected in the results, although the differences are not proportional and are not consistent, simply because the overlap among the three databases is smaller than expected.

An excerpt of some of the 45 test searches represents both very broad and very narrow topics. Limiting the search to the title field provides a level playing field, regardless of whether one database has more abstracts and/or descriptors than another. For each test query, I also ran them against the basic index generated for the bibliographic citation, the index terms and the abstracts, when available.

In the case of compound terms, exact phrase searching was used. The results from PubSCIENCE reflect both the Archive and the Current subsets together. In looking at the number of results, one very important criterion must be kept in mind: While all databases have some acceptable level of duplication, PubSCIENCE has a massive amount of duplicate and triplicate records. For example, one of my test queries illustrated that only three out of seven results were unique in PubSCIENCE. Even the distortion caused by the baffling redundancy in PubSCIENCE couldn't make it beat the other two databases in terms of the number of records in all but one of my test queries.

The Software

The software of ECD and DOE ECD is, for all practical purposes, identical. And it is far more intuitive than the software used by PubSCIENCE. The database is handled as a single unit instead of being split into confusing and contradicting subsets, as in PubSCIENCE.

The query form is attractive and efficient, allowing users to quickly choose one of the important search options from the template.

For fields that have a controlled vocabulary, such as document type, a pull-down menu lists the options. Smartly, when choosing more than one document type from the pull-down menu, the search terms are concatenated together automatically to form an OR statement. It would have been nice if there was a pull-down list for other fields, like journal name, as well.

The results are presented effectively in a matrix layout that can be sorted by one of many criteria. This is perfect when large numbers of results are returned. Again, journal names could be an important criteria for sorting.

For the results, you may choose to display only the number of hits or the hits and a short entry. If you choose the former, you can click on a button later to display the full results lists. This is ideal for searches with a presumably large number of hits (as in some of my test queries). For example, you may just want to get a feel for the depth of coverage on various topics, or see the number of hits retrieved by your search strategy.

As for the full output, it is much better in the ECD databases than in PubSCIENCE. The sample output, which also shows the availability of the full document (a fact not mentioned in the very limited PR materials), is very easy to read and is appealingly laid out. The source documents are usually of very good quality as well.

While there are many documents in PubSCIENCE that are unique, the best solution for the future is to transfer its unique content to ECD, without the crippling duplication, and focus all the energy to enhancing the ECD file.