Reference Reviews

Péter's Digital Reference Shelf

October 2006

Title: Library, Information Science & Technology Abstracts (LISTA)
Publisher: EBSCO Information Services
URL: http://www.libraryresearch.com
Cost: free
Tested: from November, 2005 onward

The Context

EBSCO has been generous to offer two open-access indexing and abstracting databases, Teacher Reference Center (TRC), and Library Reference Center (LRC). TRC has been enhanced and, as of mid-September, offers more than 1 million indexing/abstracting records. LRC also was much enhanced then replaced last November by LISTA, which includes also at least 5,000 records from the Information Science and Technology Abstracts (ISTA) database acquired by EBSCO a few years ago.

For your background, ISTA is a cleaned up version of the now defunct Information Science Abstracts (ISA) database [isa-1] which has been ill-treated by its content producer, IFI/Plenum Data Corporation for more than 15 years since the early 1980s. Details of the neglect and its consequences have been provided in various publications by this author, as a memento of the embarrassing history of a database of our profession. Interestingly, EBSCO still offers the subscription-based ISTA database, but its days are likely numbered. It is hard to imagine that anyone would pay for a database that offers merely 15% of what the LISTA offers free.

While TRC is not considered to be a breakthrough (simply because of the very comparable open-access version of the ERIC database), LISTA is a much welcome novelty. There are tens of thousands of indexing/abstracting records related to library and information science and technology in ERIC, PubMed, NTIS, and some of the other government databases, but LISTA dwarfs all of these together in size and many other regards discussed below.

The Content

Size

LISTA had 957,286 records as of mid-September, and keeps increasing every day. This makes it by far the largest among all the commercial indexing and abstracting databases specializing in library and information science and technology. For comparison, ISTA has about 170,400 records, LISA (Library and Information Science Abstracts) of CSA has 290,000, LibLit (Library Literature and Information Science) of the H.W. Wilson Company has 293,860. (H.W. Wilson also offers a huge retrospective edition of the database covering the 1905-1983 period with about 500,000 indexing-only records. Another edition features full text articles from a group of select LIS journals).

Of the multidisciplinary databases that clearly identify subject areas for library and information science, the T&I (Trade & Industry) database of Thomson Gale has 296,500 records in the Library and Information Science category, while SocSCI (Social Science Citation Index of Thomson ISI) has 215,3000 records for the Information Science and Library Science category).

Composition

It is to be noted that an unusually large proportion, 58% of the LISTA records, are for book reviews of all types of books in terms of genres and disciplines. Only SocSCI has a similarly large proportion (54%). In LibLit, the ratio is about 6%. LISA doesn’t provide this kind of breakdown by document type. The positive aspect of the large proportion of book review records is that it offers a useful collection development tool. It very much sweetens the deal that many book review records are linked to the full text of the reviews in other, subscription-based databases of EBSCO.

This leaves more than 350,000 records in LISTA for feature articles, editorial materials, books, book chapters, conference papers, database and other product and service reviews.

Journal base

LISTA has a broad journal base covering about 700 journals and other periodical publications. About 425 periodical publications are designated by EBSCO as core journals of the LIS&T literature, and 165 publications as priority journals. There are about 100 more journals with very selective coverage (that’s why EBSCO refers only to 600+ journals in its database description.

ISTA, LISA and LibLit have a narrower base of 400-450 periodical publications. Of course, wider coverage alone would not be enough, and LISTA (for most of the journals) does not have as long retrospective coverage as LISA, LibLiT and SocSCI. The time span of the coverage can be analyzed on a journal by journal level. It is a very arduous job, so I took some samples, using traditional and new journals, ones focusing on library science, others with emphasis on information technology.

The depth of coverage of LISTA (in spite of its shorter time span of retrospective coverage) seemed to be good or very good in my sample searches – with one notable exception, the Annual Review of Information Science and Technology (ARIST), one of the top three ranked serial publications by impact factor in our field. The 102 hits for this publication may look good in comparison with ISTA (94), and LISA (0), but not with LibLit (207), and especially not with SocSCI (400). Considering that ARIST is in its 40 th volume, and on the average there are 12 topical chapters in every volume, SocSCI with its 400 hits is the only one with good coverage.

I think the very erratic coverage of ARIST is the curse of the ISA heritage which was not fixed in ISTA, either. It seems from the records that EBSCO took the records for ARIST (and about 5,000 other records, especially records for patents), from the ISA/ISTA database (except for the most current volumes), trusting that one of the most prominent sources would be well covered by ISA. It wasn’t, and neither were many of the LIS journals identified as core journals in the appendix of the print version of ISA, but not treated like that in reality.

On the upside, LISTA has 3,734 records for Online (searched in LISTA by its ISSN to separate it from other journals which have the word ‘online’ in their names). Limiting the search to ISSN is fairly reliable in LISTA as 93% of the records have ISSN. Only T&I had more (actually significantly more) records for this core journal (5,556) while the others had fewer, often far fewer records: ISTA (1,554), LISA (1,908), LibLit (2,239). SocSCI is very comparable to LISTA with 3,483 hits, (but of course it has cited references for 82% of its records – an increasingly important consideration).

For Library Trends, the hits were as follows: LISTA (1.325), ISTA (1,117), LISA (1,406), (LibLit (1,001), SocSCI (1,491). LISA and SocSCI hade more hits than LISTA for Scientometrics, and Libri LibLit also scored higher than LISTA for Online Information Review . Not surprisingly, ISTA had most often the lowest hit counts for my test periodicals, even for web-born journals, such as First Monday (which were obviously not the “heritage” of the ISA database from the IFI/Plenum management era).

Subject coverage

PR materials of most databases list a wide variety of topics, but these should not be taken for granted. In test searches – limited to the title field for level playing field comparison – LISTA scored the highest not only for hot topics, but also for traditional LIS topics. For example, the query for the phrase ‘open access’ in the title field LISTA yielded 340 hits compared to ISTA (62), LISA (253), LibLit (192), and SocSCI (218). The query ‘bibliographic databases’ showed quite comparable pattern: LISTA (126), ISTA (112), LISA (148), LibLit (115), SocSCI (105). For the term ‘citation analysis’, not surprisingly, SocSCI had the highest number of hits (349), followed by LISA (265), and LISTA (211), while ISTA (165) and LibLit (162) were neck to neck. These subject searches in SocSCI were not restricted to the Information and Library Science topical domain.

The good coverage of academic and professional journals and other sources of LISTA is indirectly demonstrated also by the results of the search for Carol Tenopir as author who has been publishing intensively in both types of publications: LISTA (334), ISTA (174), LISA (304), LibLit (322), SocSCI (315).

Currency

While the retrospective coverage of LISTA is not as good as in LISA, LibLit and SocSCI, it shines when we look at the other end of the time line, and examine how current is the database. Usually there is a time gap of several months after an article is published in print and an indexing and abstracting record of it becomes available in a database. Many years ago I tested the currency of some databases, and found the gap to be from 4 to 9 months.

In LISTA (and many other EBSCO databases) there are records for thousands of articles which have not yet been published in print. Apparently, EBSCO made arrangements for a direct feed from some of the largest scholarly publishers. That is the only explanation how could be there records for articles to be published in print in the last quarter of 2006, and even in the first quarter of 2007. These are not merely puny pre-publication records. Although there are no descriptors assigned to them, this is more than compensated by the substantial abstracts from Springer, and Elsevier 2007. As will be discussed below, for some of the users this currency is topped by the availability of the full text document, fetched by EBSCO from the publishers’ site.

Record content

LISTA has abstracts for 937,878 of its records (78.5%). True, some of the abstracts, especially the ones for book reviews, add very little (if any) to the information provided already by the other data elements, - a common weakness also in many databases of other content producers which include records for book reviews. However, quite a large proportion of abstracts, especially the ones which are directly taken from scholarly journals are substantial and informative. This proportion of records with abstracts looks good, but both ISTA (90%) and LISA (93%) do better. LibLit has no abstracts (except for about 3% of the records).

About 83% of the records list the authors. This is a tad lower proportion than the usual 90-92% rate in other databases , but it is mostly caused by the large number of unsigned news items and very short editorial materials. This in turn, is the consequence of comprehensive coverage of the core journals in LISTA – at least for the for the past few years.

As for completeness of the records, there was one problem area. Visually scanning many result lists indicated that even longer articles missed the name of author(s). It is impossible to estimate how many records omit the author names where they are identified in the source and so should appear in the record. Take as an example the case of the first two records in the result list of 8 hits on the topic of citation searching which seemed very familiar to this un-credited author.

Equally frustrating are the misspelled author names, which showed up with much higher rate in LISTA than in any of the other LIS databases. It was not uplifting to see my name misspelled in 37 records, almost 10% compared to records with my name spelled correctly.

A small proportion of the records for recent articles in various result lists appeared with a very useful set of value added data elements, a list of cited references. Records of about 170 of the core scholarly journals in LISTA have been enhanced by cited references – although only for the past few years at best. In spite of the small number of records (about 5,700 in my test), this is a very promising move.

Much more importantly (for a certain type of users), more than half of the records are linked to the full text document in HTML and/or PDF format. This took my breath away as we are talking about 552,560 documents, including about 300,000 substantial ones, not just news items. This works only when using the database from the library or other campus locations directly with IP authentication, (or logging in through a proxy server remotely), otherwise this feature does not show up in the result lists.

However, if LISTA is added to the EBSCOHOST profile by the system administrator in a library, the full-text documents become available, and this is clearly indicated in the result list. Apparently, they come from several full text EBSCO databases (Academic Search, MasterFile Premier, Business Source, Computer Source) to which the library subscribes.

This is a great idea because most of the users would not hop from one database to the other and run the same query in hope of finding full text articles. This is done by the software automatically, behind the scene, and instantly – a perfect example for convenient autonomous resource discovery beyond the borders of the database being searched.

In addition, if the library uses a link resolver software, many of the full text documents can be retrieved from the digital archives of several of the largest scholarly publishers (such as Elsevier and Springer) to which the library subscribes. The rate of successful look up of full text documents depends also on the time frame of the subscription by the library, and it is not predictable when looking at the result list, but as we shall see it is worth a try.

The Software

The standard user interface of EBSCO is used with the LISTA database, so you would see the latest features deployed, such as the visual interface powered by the Grokker software, and the descriptor list in the sidebar. There is a generous selection of browsable indexes, which provides the posting information of the index terms, showing in how many records they occur.

There is a basic and an advanced search mode. The latter offers a number of check-box and pull-down filters to limit the search to records with links to full text documents, peer reviewed journals, a time period, a specific journal, and/or to document types. There is a third search mode for citation searching, which is laudable, and is very flexible.

Two of the check-box filters do not work. One is meant to limit the search to cover stories (a very useful option to limit the search to articles which are one of the main themes of an issue of a journal), but as no such designation is used in LISTA for the documents, the result will always be zero.

The other check-box also has great potential to limit the search to records which have been enhanced by a list of cited references, as these can lead the users to records about closely related other articles. For example, in the result list of a highly specific search on the topic of the financing of open access publishing, the record about an article published in mid-September, 2006 and enhanced by 26 cited references can be particularly useful for having a good bibliography.

Just knowing this number in itself is a valuable and useful hint, but in LISTA, one can display the list of cited references. (The list also shows how many times the items on the cited list were cited – but this is a weak point in EBSCO).

Unfortunately, the check-box approach does not work. However, you can achieve the same function (in a far less apparent way) by adding the FR Y filter to the topical query.

The link-out label appears only when logged in to a system running a link resolver. Even then it merely shows a chance to be linked to the full text article hosted by the publisher or by other third party. Regular users will rather quickly acquire the knowledge which journals in the result list have a digital archive available, and they will enjoy this cross linking. In this example the user is authenticated by the third party computer hosting the digital edition of the journal, which then informs the user that the article is available for download for free in PDF format.

As mentioned above, a message with citedness count appears in some records below the title of many articles. You would rarely find a citedness count larger than 1, and most often you don’t see any citedness count because 0 citedness is not reported (understandably). For the above article, however, there are 5 citations in the LISTA database with various errors in the name, the journal title and other elements. The matching algorithm of EBSCO often turns out to be too strict. It is the opposite extreme of what Google Scholar is doing which often reporting a cited/citing item match when none exists as illustrated in an earlier conference presentation, and in a paper about deflated, inflated and phantom citation counts .

This review is not the appropriate place to go into details about this highly technical issue. Suffice it to say here that EBSCO creates a character string from various elements of the bibliographic description of the source items and from the cited references. The string is different even in case of differences of lesser importance, such as when the second initial of the third author is present in the cited record but absent in the citing record. Given the excessive volume of errors in many of the elements in citations, this ultra-orthodox matching requirement will cause considerable underreporting and should be re-tuned to find a compromise between the extremes. The good news is that EBSCO offers a very flexible search option which helps in discovering also the quasi matching citing items in the references.

In spite of some deficiencies, this is an exceptionally useful database for librarians and information professionals who have now open access to the largest and most current indexing/abstracting database on library/information science and technology, as well as to half a million full text articles (if they subscribe to the most commonly licensed EBSCO databases). For those libraries which have a link resolver installed it is also a perfect tool for bringing out the most from the publishers’ digital archives subscribed to by the libraries. Do yourself a favor, and have EBSCO add LISTA to the menu of EBSCO databases for maximum deployment, or bookmark the site http://libraryresearch.com if you don’t subscribe.

— Péter Jacsó