PIE Import/Export Functionality

Required functionality

It would be desirable for the PIE to be able to export and import data into a variety of formats. These include:

• Export lists of resources to Netscape bookmark file - this would be required to add a new folder to an existing bookmark file
• Import Netscape bookmark file into a page - this would be required to retain the folder structure as lists in the page
• Export lists of resources to Internet Explorer fluorites' folder - this would be required to do this as a new sub-folder
• Import Internet Explorer favorites' folder into a page - this would be required to retain the folder structure as lists in the page
• Export search results into a format which can then be imported into bibliographic software e.g. Endnote

Issues

• When exporting to bookmark file or favorites' folder, it is vital for the software to check for conflicts between new folder titles and existing ones. (It is probably best to export to a new folder rather than attempting to merge with existing ones.)
• There are security issues with any attempt by the server to write to a user's disks, which need to be bypassed. This means that most of the available bookmark utilities rely strongly on the installation of software on the user's machine.
• Institutions may well have tackled problems due to use of shared machines in different ways, so that bookmarks files and favorites' folders may be in non-default locations (e.g. on the LSE campus, most users store bookmarks on a network drive). This means that getting the user to install software is a poor option; many users will be unable to do this.
• For exporting search results into a format understood by bibliographic software, it is important to find one which is as generic as possible, even if this means a loss of information in some cases, because then the system is as future proofed as possible and will also be more likely to be extensible to new software if the PIE is used by someone who uses a radically different bibliographic tool. (If there is an open standard format, that is probably well worth using.)

Possible solutions - bookmarks

The above considerations make it look as though the only coding options are going to be Java or Javascript, to provide the required level of integration between the user's local environment and the browser. Checking Freshmeat (Linux applications) and CPAN (perl modules) reveals nothing appropriate.

Internet Explorer uses a Javascript extension, the function

ernal.AddFavorite(url,title)

to add a URL to the main folder of favorites; Netscape has no equivalent. An example script to add a page to the Explorer favorites folder can be found at the Javascripts archive.

However, the simplest solution to implement for importing/exporting both bookmarks and favorites is to use the existing import/export tools already present in the browser. Netscape can import the links from any HTML file; they are placed into a new folder in the bookmarks file, with the folder title given by the HTML file title, and the links organised into sub-folders by sections given by header tags in the document. The file needs to be local and so the user will need to save it. The file format of the Netscape bookmarks file is easy to replicate (and is now license free).

There are several Windows tools to import Netscape bookmarks into Internet Explorer favorites folders; Microsoft recommends this one. Internet Explorer, from at least version 5.0, will export favorites as a Netscape bookmark file. Navex, however, will import and export in both directions.

Possible solutions - bibliographic software

Endnote has a mechanism whereby information providers can submit what ISI call "import filters". In order to import data from an information service, the user has to install the appropriate import filter. There are several points to note:

• Endnote can automatically connect to any Z39.50 bibliographic database, do a search and save the results in its internal format. I would expect many users to do this rather than import from a third party portal service. The LSE library provides users with some preset filter files (e.g. for the library catalogue), which are part of the school standard build.
• There are over 200 free to download connection profiles for the common online services, which means that the user doesn't even need to know anything about the Z database to which the connection is made.
• The import filters are in a proprietary format.
• Procite offers essentially the same functionality, though in this case the filters are configured as part of program set up.

Endnote is also able to import from a variety of ASCII formats. However, it principally uses a modification of the refer format (developed from that used by the Unix troff document formatting software). (The main differences are some extra tags and a slightly different meaning to some of the standard tags.) A perl module, bp - by the same author as this documentation - is available to convert between a variety of bibliographic formats including both straight refer and the Endnote version.

A comparison of Endnote with other bibliographic software - with particular reference to supported import and export formats is available from the Berkeley University academic writing programme. All of the considered options except ProCite will import from the Endnote export format described above.

Conclusion

While for more the technically minded, the best option is likely to be to use Endnote's ability to directly get bibliographic details from Z39.50 searching, I would suspect that most users would find this beyond their capabilities. It does seem that the Endnote format is likely to be the best bibliographic database format, and the bp module is probably the best way to achieve not just this but a degree of interoperability with the variety of formats it offers.