ECHET96 Search CD [Molecules: None] [Related articles/posters: None]

HyperWave: An Electronic Conference Manager

Omer Casher and Henry S. Rzepa

Department of Chemistry, Imperial College, London SW7 2AY.

HyperWave, often referred to as Hyper-G, is a sophisticated "hypermedia" system that is used extensively in both academia and industry. Designed to handle large volumes of documents distributed across multiple Web sites, it is uniquely suited for technical documentation and electronic publishing. We illustrate here why HyperWave would also be an ideal manager for the electronic conference. With careful planning HyperWave could be configured to run an entire electronic conference automatically. Moreover, its multi media support and recently developed applications could, in fact, propel the electronic conference into entirely novel domains.


Limitation of Conventional WWW Servers

Accessing ECTOC or ECHET96 documents on the Web is not unlike accessing most other documents. That is, following the request from the client browser, the WWW server retrieves the documents from a directory hierarchy and sends them to the client via the HTTP protocol. Such a process is adequate for a small collection of documents. However, given the large volume of documents within an electronic conference, many of which are repeatedly updated by several users, document management and the maintenance of hyperlink consistency are arduous especially across multiple servers. Additional tasks such as CD-ROM preparation, and the processing of molecular structures associated with each paper still require considerable human intervention.

The Molecular Hyperglossary [1], linkdb, and other in house packages were developed to assist in electronic conference management. Although they largely circumvented WWW Server limitations, it was clear that to fully address document management issues, an alternative to the WWW Server itself was needed. HyperWave's [2] potential for electronic conferences was clearly evident once we implemented it in an electronic journal [3]. Enhanced articles of Chemical Communication are now accessible on a HyperWave Server.


Document Management

All HyperWave documents are members of document collections (Fig. 1) and a document can be a member of one or more collections. A conference scenario might have all papers grouped in a single collection allowing users to view a list of the entire conference content if they choose. Papers could also belong to different collections based on subject matter. Most likely, each paper would itself have its own separate collection.

Figure 1: The collection of keynote articles

The indexing of documents permits rapid searching against title, keywords, name, or body. For now, only text in HTML or HyperWave's own HTF format are indexable against an appropriate DTD. Manuscripts submitted to the conference as HTML could therefore be inserted into the server by the author and indexed immediately. In the next release, indexing against user specified metadata will be possible. Each HyperWave document and collection has associated metadata (Fig. 2) and user defined metadata is permissable. Selected meta attributes could be indexed and individually searched. A filter is under development to index PostScript and filters for PDF, Microsoft Word and VRML documents are planned. User created filters for other document types, such as SGML with specialized DTDs, will also be possible.

Figure 2: The metadata of a PDB file

A variety of these filters would allow manuscript submission and modification in formats more familiar to the author. Documents would therefore need to be kept on the server in their native formats. The rendering of non HTML (or VRML for 3D models) in the client browser would likely require a suitable on-the-fly file conversion into HTML (or VRML). As both the HTML and VRML formats are evolving, creating these documents dynamically is advantageous. HyperWave's document cache server precludes the need for repetitive conversions.

In ECHET96, manuscript submission in HTML via a forms interface was encouraged. HyperWave has a similar document (Fig. 3) submission mechanism. In cases where submission of manuscripts, referee reports, or user comments by e-mail is more desirable, HyperWave's mail tool could be used to manage them appropriately.

Figure 3: Text insertion via a Netscape form

Document read and/or right accesses can be restricted to a single user or groups of users. A user can belong to more than one group. Time constraints can be set where a document can go "live" at a certain time and become inaccessible at a subsequent time.


HyperWave Administration

HyperWave servers are administered either by a suite of server side modules or by a HyperWave client. Harmony and Amadeus are authoring environments for the UNIX and PC respectively and are similar in functionality. The contents of a HyperWave server can be viewed as a hierarchy tree resembling the UNIX file system. Searching and document manipulation are carried out via pulldown menus. Client/server communication is handled by HyperWave's own stateful protocol.

Usually HyperWave administration is shared by a main administrator and trusted users who have read and perhaps write privileges for a subcollection of documents. In a conference setting, each author could be given a personal collection to add, remove and modify documents at will through a forms driven interface. In case of errors during modifications provisions to keep a replication of individual papers should be in place (see Conference Mirroring)

Harmony and Amadeus are being largely superseded by an HTML forms interface and a Java based client that promises to incorporate drag and drop features for document manipulation. This application has significant implications for the electronic conference. In the future, a conference could be coordinated through any Java aware Web browser. By this same means authors would be able to readily submit and organize the contents of their own papers. Furthermore, the complete conference can be browsed through a dynamically generated hierarchy tree running as a Java applet thereby precluding the need to maintain a table of contents.


Hyperlink Maintenance

HyperWave separates links from documents and stores them in a link database. The links themselves have meta attributes such as access rights and expiration dates. They are usually bidirectional; that is, a document knows not only where its links point but also which documents point to it (Fig. 4). Link consistency is therefore guaranteed. If a document is removed, all links to it subsequently vanish from all HyperWave servers. Provided the links themselves are not removed they will reappear if the document is replaced.

As conference papers are often moved, removed, replaced or modified such link maintenance is vital; more so if the conference were spread over several sites with each site maintaining different documents. Of course the remote sites would have to each be running a HyperWave server, which might be superfluous for a site maintaining just its own conference paper.

Hyperlink maintenance is not restricted to HTML but also extends to PostScript, PDF, VRML, images, MPEG movies, and audio files. Conference papers, or even sessions, supplemented with voice and movie annotations can be envisioned.

Figure 4: All hyperlinks pointing to and from a PDB molecule


Dynamic Documents

HyperWave's support of document templates allow features common to several documents, such as pushbutton images or Java applets, to be contained in a separate template file and assembled with the rest of the document when accessed. The clear advantage is that any changes to the common features need only be made once and the changes will appear on all documents containing the template file. This feature would be useful in making full articles and posters more distinguishable by giving each a common look and feel.

Document templates can also be used to construct tables or glossaries dynamically. Whenever a new paper is added related tables would be automatically appended with no human intervention. The table depicted (Fig. 5) could easily be constructed dynamically once the HyperWave server interrogates the meta attributes of each document in the Stereocontrolled Reaction collection.

Document templates also allow multiple document clustering where one HTML page is a composite of smaller documents. The reader would therefore have the choice of viewing a paper as one complete document or toggling through several smaller documents with the help of a glossary.

Figure 5: A typical table of conference papers


Conference Mirroring

Whether a conference is administered at one central site and mirrored to other sites or administered at several sites each mirroring the others, keeping all documents at all sites current is tedious. As with mirroring any web site, the content is most likely transferred in bulk, which is time consuming, largely superfluous, and a hindrance to network bandwith. Ideally, only the modifications to individual documents should automatically propagate to the other sites once they have been made. This would necessitate using a so called mirroring daemon.

HyperWave is developing a mirroring daemon but its excellent document export and import tools do allow for fine tuning of site mirroring, albeit without instantaneous propagation of changes. This is achieved by querying the meta data of all documents in a specified collection and mirroring only those documents that match the query. Conference mirroring can, for example, be restricted to documents of individual papers or to documents created or modified after a certain date. Before a document at a mirror site is overwritten, various "safety checks" can be made such as assuring that the document is actually older than the one being mirrored. Mirroring can be configured to run automatically at specific times. For slow networks a timeout option permits up to five attempts at mirroring a document before moving on to the next one.


CD-ROM Production

Browsing conference proceedings from CD-ROM should be analagous to browsing local files. However, the initial process of transferring the conference on to CD-ROM is again lengthy, requiring the organization and verification of documents and their hyperlinks. As the CD-ROM version stands as a permanent record, all text must strictly adhere to HTML rules so as to assure longevity.

HyperWave has a standalone server component which, with the help of a HyperWave client, accesses a "local database" without connecting to a HyperWave server. As with a normal HyperWave database the local database contains indexed searchable documents. Because HyperWave documents are always assured of hyperlink consistency and HTML integrity, archiving an electronic conference as a local database on a CD-ROM is a highly efficient approach. HyperWave's CD tool is used for this purpose. A recently developed Java applet can be used to browse the documents as a tree hierarchy. A demonstration of this can be viewed here.


Database Integration

In addition to being text-based searchable, conference proceedings should also be structure and substructure searchable. ECTOC and ECHET96 provide a Hyperglossary [1] for depositing structural data associated with a given article. Java applets for molecular editing are now available and, if used within a hyperglossary, will facilitate structure creation. The database's own search engine can then be incorporated into the conference's search facility. Ideally, hyperglossary input should be automatically inserted into a database without human intervention. Once inserted the data should be readily retrievable to be incorporated, with the help of suitable rendering tools, either into the corresponding article or as part of a query result encoded in HTML.

HyperWave is able to access an SQL database using its own gateway interface, HGI. A common gateway interface (CGI) request which initiates the execution of a program and its subsequent termination once the request has been processed, is ill-suited for a legacy database. Unlike CGI, HGI maintains a persistent connection to the database and does so tarnsparently. That is, an entire set of SQL queries could be organized as a collection hierarchy within a HyperWave Server each having its own meta data. Depositing and accessing structural information from a Web frowser to and from a database could therefore be made to appear totally seamless.


References

  1. C. Leach, P. Murray-Rust and H. S. Rzepa, "A Molecular Hyperglossary: Organic Molecular information in Hypermedia Form", Elec. Conf. Tr. Org. Chem. (ECTOC), 1995.
  2. H. Maurer et. al."HyperWave, The Next Generation Web Solution", Addison-Wesley, Harlow, UK, 1996.
  3. H. S. Rzepa, O. Casher and B J. Whitaker, "A Paradigm Shift in Chemistry Electronic Publishing", Proceedings of the 1996 International Chemical Information Conference, Nimes, October 1996. ISBN 0-952-1097-43.
  4. J.H. Kreiger, "Doing Chemistry in a Virtual World", Chemical and Engineering News, 74, 50, 1996.