 |
|
|
| ||||||||||||||||||
|
|
|
| ||||||||||||||||||
|
|
| ||||||||||||||||||||||||
| ||||||||||||||||||||||||||||||||
|
 |
|
|
 |
 |
|
|
|
|
|
|
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
The Synergy of Special Databases " . . . the aggregate effect of the interaction of Dialog power
commands and special databases with the various databases provided by
third parties is greater than the sum of individual effects of the
components Introduction It is common knowledge that people are most likely to count, propose marriage, and pray in their mother tongue, no matter how long they have been living in a different culture. My mother tongue of online searching is Dialog. While I enjoy the gracious graphical thesaurus implementation of Ovid, the user-friendly sort and print specification of WinSPIRS, the smart result ranking of KnowledgeFinder, I always switch back to Dialog when a search requires the power commands of Dialog (MAP, RANK, EXS, REPORT), OneSearch® for cross database searching, or one of the special Dialog databases like DIALINDEX®, or Journal Name Finder™. Some other online services may have one or two of these features, but none has the line-up of Dialog. When these Dialog power features are combined for a particular assignment, the savings over traditional methods of getting information can be very significant.Tinkering with scripts for DialogLink®, RANKing or MAPping the result of a query, EX-ing the TEMPSAVEd result in OneSearch, and REPORTing the hits of a Journal Name Finder search is not unlike the best action of the (first) Dream Team: David Robinson blocking a shot, passing the ball to Drexel, who fakes a pump, breezes to the right and finds Magic with the ball who assists Jordan for a dunk with a no-look pass. My search story has to do with coaching the dream team of Dialog power commands and special databases in an exceptionally demanding situation that called for all of my online searching expertise — and proved the synergetic effect of Dialog advanced commands and special databases. It was like an exhibition game.
The Exhibition Game The New York-based Soros Foundations decided to donate computers with CD-ROM jukeboxes and to subscribe to the most relevant 10-12 databases for 25 medical information centers in 13 countries of Eastern Europe, the former Soviet Union, and the former Yugoslavia. I was the consultant in the project to recommend the hardware configuration, the mix of bibliographic and full-text databases, (and to provide the training for the users in a one-week course in Budapest).Obviously, the database recommendations for the primary and secondary literature had to based on the information needs of doctors, medical researchers, and faculty members in the recipient countries. The text-book strategy of analyzing the information needs of the users through fact-finding missions, personal interviews, or mailed questionnaires was not feasible for several reasons. Organizing a trip from Albania to Croatia through the Baltic states would have been not only very time-consuming and expensive but would have made even the most seasoned travel agent quit her job. At the beginning of the summer vacation season, faxing or mailing survey questionnaires, responses, and their clarifications back and forth would have been also unrealistic. The challenge was to find out which are the most appropriate databases for the intended users in this project. The answer was Dialog for the reasons described below.
The Game Plan The idea was to get a profile of the users' preferences for medical literature in the 13 countries - indirectly. The strategy was to identify
You will not find current and systematically analyzed reports about these issues in your neighborhood library, not even at NLM. However, the large collection of databases on Dialog, along with the powerful set of Dialog commands, could deliver the best answers with some coaching. The game plan called for a series of searches that were scripted, tested, and automatically re-run for each country (changing only the country name and language) in absolutely consistent fashion, in a nick of time. DialogLink allowed preparing, editing, and saving the queries; capturing RANK lists and reports; tinkering with and resubmitting saved searches generated by RANK and MAP; and resubmitting them at modem speed instead of the error-prone and slow re-typing of each query. Once the raw data were available through appropriate searches, they were converted into information, then into knowledge, using the power features of Dialog described below — in a few days and for almost peanuts. (At that time RANK had no surcharge, but connect-time charges were higher than now). Starting the Ball Rolling The following cross-database search in EMBASE®, MEDLINE®, Life Sciences Collection, and SciSearch® yielded the number of records for each country (the example shows Hungary). The latter being an interdisciplinary database required to limit the search to the various subject categories of medicine, and the use of the GL= field for country of publication. The timeframe was limited to 1994-95, but was extended in case of too few records for other countries. It was understood that journals (edited and) published in a country may have many foreign authors, but this represented a small percentage in most of the cases, and was offset by the advantage of learning about the coverage of journals published in the recipient countries. The set is RANKed by journal name, and the top 200 journals where the authors published are SearchSaved. (See Sample Search, Figure 1.)
Full Court Press This set is executed in Dialog Journal Name Finder (File 414). Dialog generates a record in this database every month for the JN and JO indexes of those databases that use this field. The record also includes the number of items, the record count (RC) from that journal (in the given database). The resulting list of the search in File 414 shows how many databases cover each of the journals stored in the SearchSave (only the top 5 are shown here). The extent of coverage of a journal cannot be seen from this result but we shall get there. First, however, the set is reduced by filtering out those journals that have a single item (See Sample Search, Figure 2.)Fake Pump The extra beauty of File 414 is that it creates both word and phrase indexes (with and without prefix) from the content of the JN and JO fields. When the resulting set of the SearchSave is ranked by journal name it lists — without a prefix — the journal names as searched and also the different spelling variants that it found. It is nice, but this ranking step was just a fake pump to show what happens behind the scene, and to be ready for the no-look pass. (See Sample Search, Figure 3.)
The No-Look Pass The parameters of the MAP command can deliver some magic. For example, in this case we want to have the journal names mapped without a prefix (that is the /= symbol). Similarly, suffixes can be added, and prefixes can be replaced. I used this latter strategy to convert the CW prefix into the JN prefix when looking for the most cited journals of authors from these countries (not discussed here for space limitations).The MAP command with these parameters creates a query with positional operators based on the spelling of the journal, including () and (1w) to take care of the space character, the dot, comma, hyphen, as well as of the stop words, such as 'and', 'of' etc., and executes the query immediately (exs parameter). Again, only the first few entries are shown. (See Sample Search, Figure 4.) The Dunk After the execution of the mapped SearchSave the journals with single hits (mostly misspellings) are eliminated, and so are the archive files that would be redundant. The reduced set is then ranked by file number (FI). This step usually reduced the set by 40-50%. The resulting list shows how many journals are covered by the qualifying databases. Again, only the first eight databases are shown for this small demo set. Unfortunately, they are identified by file number instead of file name. (See Sample Search, Figure 5.)
The Extra Free Throw If you wish to know the details about the coverage in each database, sort the set by journal name within database number, and create a report including the file name, the record count and the journal name. Alternatively, you may sort by journal name as a primary, and record count as a secondary sort criteria, and create a report in that arrangement for further analysis. (See Sample Search, Figure 6.)
Overtime The steps in this search were followed by comparing the potential databases and full-text journals on CD-ROM in terms of retrospective coverage, software capabilities, interface features, and price, but it is a different story for a different essay competition. Suffice it to say here that this was a challenging assignment that brought out the best of Dialog (and perhaps of me, too), provided a lot of useful and consistently structured information, and cost much less than a traditional survey that is hard to reproduce, and may be far less objective. This exercise proved that the aggregate effect of the interaction of Dialog power commands and special databases with the various databases provided by third parties is greater than the sum of the individual effects of the components of the service...just like in the Dream Team. (See Sample Search, Figure 7.)
Afterplay The medical centers have been happily using the databases, and now want to extend their services to online resources, including pharmaceutical ones. The Foundation is considering the financing of a new project. And I'll consider going again for such a power trip. Afterall, Dialog saved the Foundation money, and made me money. Péter Jacsó is associate professor at the library school of the University of Hawaii. He is also columnist for Database, Computers in Libraries and Information Today, and is a frequent contributor to other library and information science journals. Formerly, he won the Excellence in Writing award from UMI, and the The Electronic Library Best Paper of the Year award from Learned Information, Ltd. He can be reached at jacso@hawaii.edu.
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
 | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
