Citation Managers and Citing-Cited Data Previous   Contents   Next Issues in Science and Technology Librarianship Summer 2002 DOI:10.5062/F45M63NW URLs in this document have been updated. Links enclosed in {curly brackets} have been changed. If a replacement link was located, the new URL was added and the link is active; if a new site could not be identified, the broken link was removed. Citation Managers and Citing-Cited Data Brian Simboli Science Librarian Lehigh University brs4@lehigh.edu Min Zhang Computer Consultant Smart IMS, Inc. miz2@lehigh.edu Abstract There is an increasing trend for library resources to enable identification of cited and citing reference lists, i.e., lists of items cited by documents, as well as items that in turn cite those documents. In this article, we evaluate how two citation managers, EndNote and RefWorks, provide ways to store, display, and retrieve these lists. Then, we suggest enhancements with respect to cited and citing reference data that should be adopted by any citation management software that currently lacks them. Finally, we discuss how publishers can also do more to facilitate use of citation managers with respect to these types of data. Introduction An increasing number of electronic library resources enable one to identify not only literature relevant to one's research, but also documents that cite -- and are cited by -- that literature. By enabling users to exploit the potentially tight conceptual nexus between, for example, a journal paper, the papers it cites, and the papers that (in turn) cite it, these resources enable the building of more comprehensive bibliographies than would otherwise be possible. We refer to these as "citing/cited resources" in this article. The availability of citing/cited resources makes it useful to have efficient, easy-to-use ways to organize and display such relationships. Current citation management software ("citation managers") provide some ways to do this. These capabilities are useful, for example, to authors who track literature that cites their work or researchers who track a developing body of literature about a new topic. This paper has five sections. The first identifies and discusses a number of citing/cited resources. The second and third evaluate the current capabilities of two citation managers to store and display citing/cited relationships between documents.1 The fourth suggests enhancements that would improve the ability of citation managers to facilitate the use of citing/cited data. The fifth argues that publishers can do more to promote use of citation managers with respect to the citing and cited reference lists they provide. Building Bibliographies Using Citing/Cited Data Resources The library research guide titled "{'Cited/Citing Resources:' A Special Way to Build Bibliographies}" identifies and discusses available resources. One category of such resources are databases whose records contain not only the usual bibliographic data and indexing for a document, but also references that the document cites. We will refer to the latter as "cited reference lists."2 Cited reference lists allow one to search on a known document "x" and find documents "y," published later in time, that include document "x" in their reference lists. In turn, taking these newly discovered citing documents "y," one can perform the same operation on each of them to arrive at other documents "z," published later in time, that cite documents "y." Through such iterative bootstrapping, one builds up a bibliography. ISI has been the longstanding producer of databases providing this capability. But others, e.g., CAS's SciFinder Scholar and MathSciNet, now offer it as well. Because SciFinder Scholar offers cited references for bibliographic records added to the CAS databases from 1999 forward, and MathSciNet offers reference lists for material from January 2000,3 their ability to generate lists of citing and cited references is limited. However, as these databases grow in size over time, their usefulness in searching for citing and cited relationships will increase. Electronic journals comprise another category of resources that makes it possible to start with a known reference and find references that cite it. Examples identified on the web page are Nature, Science, and Applied Physics Letters. The growing availability of such resources creates an increasing need for efficient ways to store, organize, and display citing/cited relationships between documents. We now evaluate the capabilities of two citation managers for doing so: EndNote and RefWorks. The first software is downloaded to one's hard drive while the latter is web based. How EndNote Handles Citing/Cited Data4 EndNote enables one to download or manually input bibliographic records from library databases and then store the data in a "library." A library consists of a set of records containing data about a document (e.g., a journal article). Some of the fields, such as author, title, and year, for each record can be listed in a window that comes up when a library is opened. Clicking on a row of the list opens the corresponding full EndNote record. One can manually input data into the records that comprise a library or one can use an "import filter" or "connection file" to draw data from databases into EndNote records.5 The template for the relevant reference type (such as journal article or book) determines how the data display in a record. Having provided this background about EndNote, we now discuss two of its capabilities for using citing/cited data. Storing and displaying cited and citing reference lists Recall that a cited reference list consists of references that the document cites. A citing reference list, by contrast, is a list of items that cite the document. EndNote allows one to store and view both types of list within EndNote records. One easy way to do so is manually to copy and paste such lists directly from a bibliographic database or electronic journal into an EndNote record. EndNote does not come with a distinct field for cited or citing reference lists. One can copy and paste the data into (for example) the Notes field. One can also create a reference type that includes a field labeled "cited references" and/or "citing references" and then copy and paste data into that field.6 If a given reference list is already very long, the user may decide that it clutters the EndNote record by taking up too much space.7 There are some easy remedies for this. Rather than put the data for the list into a field of the EndNote record, one can put the data into a word processor file. Then, using the link feature that EndNote provides, one can link from the record to the data. Or, one can create a second EndNote library containing cited or citing references and then in the same way link to it. However, one has to use the URL field, since the link feature is available only for this field. These methods allow one to link from several EndNote records to one word processor file or EndNote library if one wants to "cache," all in one place, cited or citing reference lists for a group of documents. In this way, a research group can maintain a single list of articles cited in the papers it publishes or, in turn, a list of references for documents that cite its papers. Doing so can be part of the continuing effort of the group to record all the background literature relevant to its current work as well as to the writing of future papers. An individual writer might also find the same capabilities useful. To summarize, one can copy and paste reference lists into EndNote records or link from the records to another file or library containing the lists. Either procedure can become very tedious if one's library contains a large number of records. It would help if cited or citing reference lists could be imported in the same way other data elements (such as author, title, abstract, etc.) are automatically imported by EndNote import filters. We tested the latter approach by using the EndNote filter for the OCLC version of PsycINFO. The filter was the latest version, downloaded from the EndNote web site. Using this filter, cited reference lists were imported into the Notes field. We modified this filter so that it would import cited reference lists into a Cited References field,8 which we added by editing the journal reference type.6 Having made these two changes, the cited reference lists were now drawn into the field labeled Cited References. However, the cited reference list wrapped, making it difficult to read the data. The example just discussed involves PsycINFO. Consider now ISI-produced data. Persons wanting to display cited reference lists from ISI Science Citation Index records, for example, will need to make changes accordingly if they want to import cited reference lists into a filed specially designed to store and display them. We have talked about how to make reference lists display in records. We turn now to how those lists can display in library windows. Recall that after one opens an EndNote library, authors, dates, and titles are displayed by default in the library window. One can change the fields that display there. For example, one can display data from the fields that contain cited and citing references lists. Doing so has only limited value, however. There is only one row of the window for each record, and so there is not enough space for much data from the cited or citing reference lists for a given record to appear. Retrieving cited and citing data Using EndNote's search engine, one can specify an author and then find all the documents in the database that cite the author. This would be a way for an author quickly to bring up records for all the papers that cite him or her -- that is, to bring up a citing reference list. The procedure for doing this is easy. After opening the search engine (available off the References menu), one then uses a drag down menu to specify the field over which to search for the author's name. If the cited reference list appears in the Notes field, one should search over the Notes field. If one has created a special field for the cited reference list, one should search over that field.9 One can narrow the search by introducing not only the author's name, but additional information such as a publication date, that identifies a specific paper by that author. To do this, use the Boolean operator "and" on the EndNote search engine. One can also do searches over whatever field contains citing reference lists. This will bring up records for example that an author (or a specific paper by an author) happens to cite -- that is, it brings up a cited reference list. This would be a way to generate a list of articles cited in the work of one's favorite author. How RefWorks Handles Citing/Cited Data Whereas EndNote is downloaded to one's hard drive, RefWorks is used in a web environment. Like EndNote, RefWorks enables one to store, within records, bibliographic and other information about documents. These records are held within "folders," which are comparable to EndNote libraries. The title, authors, and source data for journal records appear in a window that comes up when a folder is opened. Clicking on the link labeled View enables one to see the full record. One can manually input data into the records that comprise a folder or one can use an "import filter" to draw data into the records from databases. Having provided this background about RefWorks, we turn to its capabilities with respect to using citing/cited data.10 Storing and displaying cited and citing reference lists RefWorks allows users to put cited and citing reference lists into RefWorks records. One way to do so is manually to copy and paste cited and citing reference lists from a bibliographic database and then paste the data into a RefWorks record. While RefWorks does not come with a distinct field for cited or citing reference lists, five fields are available for users to insert data of any kind. A Notes field is also available. Cited reference lists can also be imported in the same way other data elements (such as author, title, abstract, etc.) are automatically imported by RefWorks import filters. However, the current import filter labeled "CAS SciFinder" imported the cited reference list from SciFinder Scholar 2001 into the "Tertiary Title" field.11 For the OCLC version of PsycINFO, the cited reference list was imported into the Notes field. In both cases the data in the lists "wrapped" and therefore it was not easy to read. Should the user decide that the cited reference list clutters a RefWorks record by taking up too much space, one can create a web page that caches the data and link to it. One can also link out from several RefWorks records to one web page, should one want to "cache" the cited or citing references for a group of documents all in one place. Retrieving cited and citing data RefWorks comes with "Quick" and "Advanced" search engines. Quick Search enables one to search all fields. Using it, one can therefore search on an author to find references that cite, or are cited by, the author. Search results are relevancy ranked. The Advanced Search feature enables one to select from a variety of fields to search. However, these fields do not include the Notes field or the five user-defined fields. Therefore, Quick Search is the only way to retrieve cited/citing lists if they appear in the Notes filed or the five user-defined fields. Citation Managers: Suggested Enhancements While EndNote and RefWorks provide ways to work with cited/citing data, their capabilities in this respect can be expanded. The following are suggested enhancements that can be incorporated not only into these two packages, but any citation managers in which they are lacking. First, a field for cited/citing reference lists could be introduced in record type templates. Vendors could also provide filters that facilitate importing of reference lists in the non-wrapping, itemized format of the original. As it stands now, recall that to import the citation data and display it in a field labeled citations, EndNote's user must edit the import filter as well as the record type template. Users may be reluctant to make these changes. RefWorks (on our trial version) does not allow users to edit filters. Second, it would be useful if references that appear in a record's cited or citing reference fields were hyperlinked to their corresponding records. Such hyperlinking would assist the user in exploiting citing/cited relationships between documents. For example, it would enable one to bring up an abstract or notes about one document and then immediately go to the abstract or notes for a citing or cited document. Also, given that one can link to the full text of a document from its corresponding record, such hyperlinking creates a network of linkages between the full text of citing and cited documents. Finally, by enabling the user to create a one-to-one correspondence between records and physical files containing offprints12, citation managers enable one to locate the hard copy of documents in one's filing system. The aforementioned hyperlinking would enable one quickly to retrieve the hard copies of citing and cited documents.13 Third, software could also perform the following function. At the user's discretion, when a record "x" is saved, the software would spawn independent records for the items in its citing and cited reference lists. This would obviate the need independently to create records for those references. An intermediate screen could be provided that would enable the user to designate which items in a cited or citing reference list should be spawned. Finally, in addition to the current summary list of records provided by EndNote and RefWorks (available when one opens a library or folder, respectively), it would be helpful to have a second view in which the citing/cited relationships between documents are displayed. For example, a document could be identified by author and date in a center column. A left-hand column would list citing documents while a right-hand column would list the cited references. So that the data do not become excessively long in that view, a hyperlink to an intermediate screen could be provided that would list the citing references, and another hyperlink would be provided for the cited references. The Role of Publishers We have discussed enhancements to citation managers that would improve their capabilities with respect to citing and cited reference lists. We now shift attention to the publisher interfaces from which users download data into citation managers. To begin with, consider electronic journals. These can be enhanced to make it easier to "harvest" citing and cited reference lists. For example, Science provides a link to "other online articles" that cite the article one is viewing (what we have called a citing reference list). Also, it enables one to download reference data for the article one is viewing to a citation manager. However, it does not enable one to download directly to a citation manager the citing reference list or the list of references that the article itself cites. Nature provides links to both "ISI articles that reference this article" and "ISI related articles." However, it too does not provide a way to download this data, nor the references the article itself cites, into citation managers. Again, Applied Physics Letters identifies "citing articles" but does not provide an interface to download them into a citation manager. Bibliographic database interfaces can also be enhanced to make it easier to download citing and cited reference lists into citation managers. Ideally, they would all enable direct downloading of citing and cited reference lists, when available, plus they would offer the ability to download to a variety of citation management software. With respect to its Web of Science product, ISI supports only direct export into EndNote, Reference Manager, and ProCite. SciFinder Scholar enables one to bring up a list of cited references or a list of citing ones. It would be helpful at that point to have a link so that one can directly download the search results in citation managers. MathSciNet similarly makes citing and cited reference lists available but also does not provide a direct downloading capability into citation managers. OCLC First Search, at least on PsycINFO, provides an export feature, but only enables downloading with it to EndNote, not other citation managers as well. In short, publishers of electronic journals and bibliographic databases can do more to facilitate use of citation managers with respect to citing and cited reference lists. Industry-wide standards might help encourage greater cooperation between publishers and producers of citation management software. We have not explored what standards in this realm have been discussed or defined. Suffice it to say that standardization in data presentation might make it easier to create import filters or direct importing mechanisms, for drawing search results into citation managers from a variety of electronic journals and bibliographic databases. Standardization may also make it unnecessary for publishers to provide separate export links for each of the citation managers on the market. Examples of standardization that could help in this context include: which bibliographic data elements (e.g., author, title, journal title, etc.) to include in citing and cited reference lists; how those data elements should appear (e.g., whether to abbreviate journal names); and what tags to use for citing and cited reference lists. Conclusion At the outset, we discussed how an increasing number of library resources enable identification not only of documents relevant to one's research, but also other documents that cite those documents. If our analysis of two citation managers is any indication, citation management software can do more to keep pace with this emerging trend. Also, publishers can explore more ways to facilitate the use of citation managers with respect to the citing and cited reference lists that they provide. Increased cooperation between citation manager producers and publishers will enable researchers to better exploit the valuable research tool that citing/cited relationships afford.14 Notes 1 Our focus is on the widely-used EndNote software, as well as another citation citation manager that has been attracting attention, RefWorks. 2 The term "reference lists" is used in MathSciNet documentation. MathSciNet uses this term to mean what we call "cited reference lists." Note that what MathSciNet calls "reference citations" are the same as what we call "citing reference lists." 3 According to CAS customer service, this {information} about the coverage of CA/CAPlus pertains to SciFinder Scholar. Re. MathSciNet, see {http://e-math.ams.org/msnhtml/whats_new.html}. 4 The paper "EndNote at Lehigh" describes how this software has been implemented at Lehigh. What follows is based on EndNote 5, specifically 5.0.2. EndNote 6 was recently released, but does not expand the functionality of EndNote 5 with respect to capabilities for handling citing/cited data. 5 See the following for an explanation of these: {http://www.endnote.com/support/helpdocs/Importing5.pdf} {http://www.endnote.com/support/helpdocs/Search_Remote5.pdf} 6 To add fields for "Cited References" and "Citing References" to a reference type -- e.g., journal articles-first choose Preference from the Edit menu in EndNote. A window labeled "EndNote Preferences" then shows up. Highlight Reference Types in the left column and click the button Modify Reference Types. Another window labeled "Edit Reference Types" shows up. Under the column for Journal Articles, add "Cited References" in the row labeled "Custom 1," and "Citing References" in the row labeled "Custom 2," respectively. Click the button labeled OK. It is necessary to do this for any record type for which one wants to store citing or cited references. This is because the generic template for record type display does not include fields for either cited or citing references. Since the available fields for any record type are drawn from those on the generic template, these fields will not be in any of the record types. 7 The problem here would almost always be cosmetic, not one having to do with limitations on the capacity of any given field to hold data: "the Notes and Abstract fields, like all other fields, can hold up to 32,000 characters which is equivalent to about 8 pages of single-spaced text." EndNote 5 manual, page 79. 8 To modify the PsycINFO filter so that it will draw in cited reference lists, first choose Import Filters/Open Filter Manager... from the File menu. Highlight the OCLC PsycINFO filter and click the button Edit. A window labeled with the filter name shows up. Highlight Template in the left column, select Journal Articles in the combo box labeled Reference Type, and add a new row for "Cited Reference:" and "Cited References:" under the columns labeled Tag and Field(s), respectively. You may save this modified filter by choosing Save As... in the File menu. 9 Note that if the user creates a new EndNote field labeled (e.g.) "cited references" for Custom 1, the search engine menu options listing the searchable fields will not call it such. Rather, it still uses the label Custom 1. 10 We relied on a trial version of RefWorks in doing our analysis. 11 The assumption here is that the "CAS SciFinder" filter can be used for SciFinder Scholar 2001 records. RefWorks indicated that the references should not go to the tertiary title field and that they can fix this. 12 For example, one of us does this by labeling with numbers file folders containing offprints, and then cross-referencing those numbers in the "label" field of corresponding EndNote records. 13 The citation manager Biblioscape has a feature very similar to the afore-mentioned hyperlinking capability. It enables one to add cited reference links to a reference and then go to those linked references by clicking. The cited links can be chosen as "bi-directional links" so that the original reference will automatically become a cross link, in turn, for the linked references. In Biblioscape, this "bi-directional link" relationship is described in the terms "cited"/ "is cited," which corresponds to what we refer to as a "cited/citing" relationship. It does not provide hyperlinks directly out of the reference itself; one must open the reference and then go to the crosslink tab to display the cited references. 14 We wish to acknowledge the assistance of Christine Capen at RefWorks for answering questions about this software and for helpful observations and comments, as well as personnel at Biblioscape, EndNote, and ISI for clarifications about their software, as well as the CAS help desk for information related to SciFinder Scholar. Thanks also to Sharon Siegler for pointing out and discussing some of the resources that appear on {http://www.lehigh.edu/library/footnote/} References ISI ResearchSoft. 1988-2001. EndNote 5 manual. Berkeley, CA: ISI ResearchSoft. Siegler, Sharon, and Simboli, Brian. 2002. EndNote at Lehigh. Issues in Science & Technology Librarianship 34 (Spring, 2002). [Online]. Available http://www.istl.org/02-spring/article4.html [August 5, 2002]. Previous   Contents   Next