jagi

Introduction   

The Web is young (Berners-Lee et al 1994a) but growing fast (see Pitkow and Recker 1994a ). Its success/failures (APWWW95 1995a) have been reported elsewhere. The content of the Web pages varies from academic home pages to research pages [turn off images] , to corporate pages [turn off images] , to very useful search engines such as WebCrawler , Lycos , SavvySearch and altavista , to the trivial and unfortunately, to the vast majority of other sites which are basically Web flotsam and jetsam often having neither form, nor content nor permanence.

What is a perceived benefit to many has been a curse to the elite few who previously saw the Net as a place for Quality information. The proliferation of browser "extensions"(Berghel 1996a) by commerce is muddying a tidy concept - uniformity of hypertext across multiple platforms. The art of proper design (Bieber and Isakowitz 1995a) is being subjugated to the need for larger, flashier in-line images.

As Ciolek (Ciolek 1996a) states

     ... Those with access (and copyright) to ample and  high

    quality   factual  and/or  scholarly materials are in the

    minority. Hence, sites will inevitably vie with each oth- 

    er for the status of being the Web's biggest (in terms of

    the catalogued hypertext links and the size of their  lo- 

    gos), or most technically advanced (in terms of the speed

    and capacity of their search engines, interactivity,  CGI

    scripts, and gateways to other software systems), or most

    colourful and dazzling (in terms of  visual  effects  and

    virtual-reality technologies).

And yet information (CACM 1992a) has become a key resource for academics, researchers and corporations. Much of our work requires communication to occur for tasks to be completed. It has been noted (Dickson and Lloyd 1992a) that seventy percent of work is on information handling tasks: hence a person's work predominantly consists of getting access to, understanding, manipulating and storing information. Access to the Internet enables multi-media communication by means of two key areas. The first is through the existing Internet mail system, the second uses its own protocol for passing information.

The e-mail standard, called Multi-Purpose Internet Mail Extensions or MIME (Borenstein and Freed 1993a) , is designed to build over the top of the current Internet mail protocols RFC821 (Postel 1982a) and RFC822 (Crocker 1982a) .

      Ever  since the inception of electronic mail, there has

    been much discussion of its even greater potential.

                                  (Borenstein, 1991)

The arrival of the WWW, HTTP and HTML allowed for delivery of text, images, sound, movies and software to any site on the Internet which had a WWW browser. This then became the standard for multimedia communication (it did however use the concepts of the MIME types and applications).

So the information needed by researchers is still there, in a more convenient format but there is too much flotsam and jetsam. The problem of "Quality Information" will be solved by "the passage of time" and by the determination and persistence of those willing to create Quality Information (see Lilley 1995a for example).

Based on previous work done(Ng 1996a, Ng 1995a) , we are currently developing a Quality Web-based journal for Computer Graphics and Computer Vision which will offer multimedia, peer reviewed articles - this free subscription journal will encourage cooperative research by enabling the easy sharing of images, algorithms and data. As part of an ongoing education programme and to encourage research collaboration, a special seminar section of the journal is aimed at knowledge sharing.

Developing a WWW-based Journal   

Most journals today present text and images. The text is linear in nature (except for a footnote or a citation - primitive forms of hyperlinks) and the images are usually of poor quality compared to what the author wants to show to the reader. The delivery of the journal is very resource intensive and is limited in distribution. The articles are often very out of date ( a turn around time of 18 months is common) and hence the articles are of historic use rather than showing what is happening now!

An electronic journal can show text in a non-linear fashion with links to other sources of information as well as present images at any desired quality (See footnote 1) ( the author can provide any level of quality limited only by the time and space available to him or her self). But also, the author can provide an audio narrative , provide the software that is being detailed, the algorithms that are being explained, the data that is being used. This can foster research in a collaborative manner rather than contribute to the "re-invent the wheel" syndrome. The author can also provide PostScript hardcopy (or the original document in proprietary format) as well as movies of results if appropriate.

Electronic Journals in Australia normally fall into the category of newsletters, etc with a few notable exceptions such as COMPLEXITY INTERNATIONAL . Worldwide, the electronic journal scene is not that different with many electronic journals actually being poor "electronic versions" of the printed journal equivalent. Few electronic journals were published earlier than 1994 (again with notable exceptions such as Journal of Artificial Intelligence Research and Psycoloquy ). Libraries (CACM 1995a) have posed and answered many questions about online journals.

Jagi originated in an aborted development project (started in 1993) using a MIME compliant email delivery system - aborted in 1994 because of the WWW which was obviously the way to go. However due to lack of financial support (first the proposal was turned down by a leading Computer Graphics company and then the proposal was left languishing for a year in a computing centre IT department) the journal had to wait until 1995 when the Electronic Publishing Working Group of The Australian Vice- Chancellors Committee awarded us a grant to set it up.

Now in 1996, e-journals are moderately commonplace - see the excellent survey of online journals at http://journals.ecs.soton.ac.uk/survey/survey.html . It is interesting to see that a number of UK-based academic publishers have announced their intention to make all their journals available online and even more importantly that the Higher Education Funding Council is supporting this initiative - take note all governments. (Take note publishers as well as there are enormous benefits in publishing online - ask any Web-based researcher.)

However the survey does not address the problem of Quality in the journals. Quality can be broken down into the following (and more) categories:

(1) Quality of content - peer review is appropriate (plus worldwide review - a factor often overlooked in criticising an e-journal - the entire Net can criticise the work and send immediate e-mail to the editor justifying their criticism).

(2) Quality of presentation - proper hyper and multi media design to give useful, informative, efficient presentation.

(3) Quality of MultiMedia - full, useful, realistic use of MM - minimise the number of 1 gigabyte inline images :?)

(4) Quality of non-linear learning - maximise the use of appropriate links without causing the user to get "lost in hyperspace".

Hancock and Hall 1995a detail the novel features of the e-journals surveyed and to a certain extent these correlate well with "quality" - they consider "searchability" (infinite cross- referencing/indexing via a search engine), "hypertext" (links to other multimedia documents), "electronic notification" (readers are informed when a new article or issue is available) and "annotation" (allows readers to make comments about the articles). (A strange novel feature was that a journal had colour photos! - what a brave new multimedia world this is!)

However, the day is young and novelty may wear off and a journal is only left with Quality to survive.

jagi   

This journal highlights Australasian research in the fields of Computer Graphics and Computer Vision. The editorial and advisory boards consist of Australasian and international researchers, recognised in their field. The journal is concerned with (but not limited to) publishing fully refereed, multi-media articles on:

 computer animation   architectural visualisation    scientific visualisation

 computer imagery     image generation algorithms    GIS/LIS

 CAD/CAM              multi-media communications     CV and pattern recognition

The journal is freely available to any researcher or organisation on the Internet network and enables the reader to receive text, PostScript documents, full colour, high resolution images, software, movies, audio and computer data. The transport mechanism is the HTTP protocol with the journal being available from our World Wide Web (WWW) site at Curtin University and mirrored from various sites throughout the world. The subscriber will "read" the journal via a WWW browser.

The journal publishes only material of high quality in the correct format. The refereeing process has the same rigour as any conventional journal. Original papers and seminars receive at least three international referee reports. Technical notes are assessed by two referees. Images are judged on quality and innovation. Each article should be an original piece of work that has not been published before (except in the form of an abstract or as part of a published lecture, review, or thesis) and is not under consideration for publication elsewhere. Each article must be approved for publication by all co-authors, and where appropriate, by the responsible authorities at the institute where the work has been carried out.

Articles are considered under the following categories: Original Papers, Technical Notes, Seminars, Images, Software and Email to Editor. The language of the journal is English. However, authors are encouraged to include a copy of all or parts of the paper in any language of their choice via hyperlinks in their title page.

Impact of the journal   

A journal of this calibre, providing these services, with world-wide circulation will have an enormous impact on Graphics and Vision research and will provide the vehicle for transporting this research well into the next century.

The feedback to the author is overnight and serious discussion can be introduced into the current issue about the article! (Since the author can be contacted via e-mail and the distribution of the journal is world-wide, the article is reviewed or refereed by the entire world rather than just by an Editorial Board!! Similarly, if the software is available, other researchers can test algorithms for validity. This raises the quality of the articles in the journal.)

In a similar manner, the contact point for an author is now no longer just an address nor even an e-mail address. Most authors will have a WWW entry on their home machine which will let people get access to more information about that person - past publications, current projects, interests, etc. It is assumed that the list of past publications may also be hyperlink documents which will let an interested reader browse and obtain other information of relevance.

Because the journal is available at your workstation, it is a part of your work. It will not be necessary to go to a library or to have it circulated (and be weeks overdue) - it is part of your information environment. This will lead to the journal being seen as an information source rather than as just late night catch-up reading! It will take its place alongside other information sources such as Archie, Veronica and, of course, a WWW browser such as Mosaic or Netscape.

Software Agents   

One of the outcomes of the Webbing of the Computer Graphics course notes was increased e- mail traffic to the authors. (E-mail from around the world asking about solving their particular problems!) This had some major educational benefits but in the context of an electronic journal, this ease of access to the journal's board members may pose serious time problems: e-mail will arrive at any time of the day (or night) due to the worldwide audience and should be answered quickly.

Hence we are looking at a Software Agent Based System (SABS) to

help process routine queries (submissions, formats, back issues, contacting authors, etc),

help route submissions to the appropriate referees (and then remind them periodically about returning their evaluation)

help with Value Added References - add hyperlinks in the reference section of submitted documents.

This System will alleviate the editorial pressure and hopefully encourage more researchers to give of their time as editors/advisers. It will also add to the quality and consistency of the journal by performing many necessary but tedious tasks.

So what is a Software Agent?

CACM 1994a gives a good introduction to the usefulness of agents as does http://www.cs.bham.ac.uk/~amw/agents/A
The agents mailing list has endless (ad nauseum) discussions re what agents are to the extent that the FAQ of the agents mailing list indicates that it "will not attempt to provide an authoritative definition...".

Agents are the subject of many articles: Franklin and Graesser 1996a and Brustoloni 1991a try to categorise agents and others, Wooldridge 1992a, Goodwin 1993a and Jennings 1993a have tried to formalise them. There are softbots (Oren Etzioni and Weld 1994a, O. Etzioni, Lesh, and Segal 1994a, Golden, Etzioni, and Weld 1994a) , robots (Brooks 1990a, Balabanovic and Nourbakhsh 1994a, Moukas and Hayes 1996a) , desktop agents (Wood 1993a, Greif 1994a) , mobile agents (Ferguson 1992a, White 1994a) (also python ) and agent languages (Shoham 1991a, Cheong 1992a, Finin et al 1994a, Thomas 1993a, Wooldridge and Jennings 1995a) (also safe-tcl and the hot new language java ).

Cheong 1996a is a good general referenced work covering agents and many related aspects of the Web however Foner 1993a gives a refreshingly insightful and amusing sociological study of an agent. It should be noted that a literature review makes apparent the fact that Agents are "hot" and that most articles could just as easily have the word "AI" substituted for the phrase "Software Agent" - Agents are good for grants :?).
(For example - http://www.cs.curtin.edu.au/documents/papers/1991/eurographics.shtml - a paper on a Visual Perception Model for a Behavioural Animation System written in 1991 could today fit nicely into the Agent niche).

For pragmatic purposes, a Software Agent will be any "intelligent" (See footnote ^2) software based assistant from e-mail filters to Web-robots.

How can Software Agents be used?   

There are a number of factors which affect the Quality of an electronic journal and it is useful to see whether an Agent can help with or takeover some of the associated tasks.

Security

Journals are often concerned with securing each edition so that it may not be accessed by non-subscribers. In our case this Access Security is not an issue.
However it is important that Transaction Security be given due attention. This may be broken down into integrity of data, authenticity of data and privacy of data.
The integrity of data can be ensured by the use of digital signatures which indicate that this document has not been altered from when the sender transmitted it. Various schemes can be used for this, notably the Rivest, Shamir and Adleman (RSA) method and the Digital Signature Standard (DSS) (criticised due to it not being seen to be at "arms length" from the American National Security Agency). The data integrity will ensure that papers submitted or referees' reports will be received uncorrupted.
A SABS may be necessary as an encrypting/signing interface to standard email systems(Weeks and Sanderson 1995a).
The authenticity of data is again important due to the possibility of academic fraud. The problems and their solutions are as above(Foner 1995a).
The privacy of email communication has been detailed elsewhere (Zimmerman 1994a, Garfinkel 1995a, Linn 1993a, Kent 1993a, Balenson 1993a) and Public Domain systems are available for use.
We are assuming that hypermedia data transmission will be secure or that systems such as mentioned in Hickman 1994a and Rescorla and Schiffman 1995a are in situ.
It is not certain if this level of security is warranted in an Internet based environment with its laissez-faire attitude to openness.

Payment

Since the journal is "free-to-air", payment is not a problem but many solutions have been proposed for subscription payments(Brands 1994a, Hallam-Baker 1995a, Rose and Borenstein 1994a, Chaum 1992a). It is obvious that Agents are not developed enough to be allowed to get involved in real transactions(Chavez and Maes 1996a).

Dead Links

It is in the area of maintaining proper hyperlinks that an SABS can excel. Many Web-traversing robots or spiders exist that will mirror sites, obtain Webpage information and check link integrity. Webtest is EIT's link Verifier robot and MOMspider also checks links and produces statistics re link failure.
Since the Web is so dynamic, the journal articles will be traversed periodically to check for dead links. When these are found, either another Agent can look for the lost page (non- trivial) or the author(s) can be emailed to determine if the dead link can be manually repaired or revived (trivial).
In a similar way, authors and their homepages can also disappear leaving old editions riddled with dead links (the electronic equivalent of a silverfish attack). An Agent could use information sources such as Netfind to try to track down the missing authors.

Notification

A user should not have to check periodically to see if a new issue has come out. Normally a user is notified of a new issue of a printed journal by it appearing on his/her desk - this is fine for static journals. An e-journal may have a static release date (or it may be a rolling monthly issue) but items may change (Letters to Editors, author amendments, etc) and the user may wish to know about these changes. Therefore the user should be notified whenever a particular section of the journal changes.
The URL-minder keeps track of Web pages and other resources on the World Wide Web, and sends you e-mail whenever your personally registered resources change. As the Url- Minder FAQ indicates:

  Q: What is the URL-minder useful for?


 A: It tells you when pages that interest you change, so you  do

 not  have to go back and check regularly yourself.  It performs

 searches on Web databases and lets you know when  anything  new

 shows  up.   It keeps track of your competition for you, and by

 running keyword searches on a regular basis it checks to see if

 any  new  competition  has  shown up.  If you have your own Web

 pages, it can keep tabs on the links in your pages and let  you

 know  when they change (for example, you can put a link in your

 page    to    the    NetMind    Free    Services    Page     at

 http://www.netmind.com and then register that URL with the URL-

 minder; when a new service shows up, you will get e-mail).   It

 lets you know when the pages you are linked to change.  It lets

 you know when users at your site change their pages.   It  lets

 you  know  if  it has difficulty retrieving your pages from the

 outside world after several tries.

This (or a similar) service can be used to help our readership.

Searching

A standard "easy to implement" feature of any online information system is the ability to search for keywords or phrases. This is especially useful for an e-journal.
Glimpse (Manber and Wu 1993a) is a very powerful indexing and query system that allows you to search through all your files very quickly. glimpse can be used by individuals for their personal file systems as well as by organizations for large data collections. Glimpse is the search engine for the Harvest Information Discovery and Access System. Harvest is an integrated set of tools to gather, extract, organize, search, cache, and replicate relevant information across the Internet. Harvest makes very efficient use of network traffic, remote servers, and disk space.
WAIS (Kahle 1989a) stands for Wide Area Information Servers, and is an architecture for a distributed information retrieval system. Like glimpse, it allows for indexed retrieval of documents - the document information can be anything: text, pictures, voice, or formatted documents.
Each issue of the journal will be indexed and a forms-based search engine interface will allow easy retrieval of information/documents.

Annotation

Annotating an article can be useful for the author as well as other readers. It can be considered as a very directed "Letter to the Editor" about that article. The annotations can be as simple as "Great" (or "Rubbish"!) or may involve hyperlinks to other sources of information (perhaps being used to support a reader's view).
The annotating of an article is simply a logistical database problem but it may have an intelligent support system similar to Hypermail to keep track of annotation threads. It should be noted that steps to minimise the well known Flame-war problem should be taken so as to ensure the continued presence of the journal on the Web :?).
This annotation can also aid in the following useful feature as well.

Collaborative Information Filtering

It is possible that a reader's interests can be ascertained after having viewed many articles. This profile could then be used to find and/or forward similar articles to the user. Similarly, if another user has a similar profile, then new articles of interest to the second user should be brought to the attention of the first. This type of Collaborative Information Filtering is described in Terry 1993a and Goldberg et al 1992a . Positive and negative annotations may be used in the selection of articles appropriate to the users profile.

Value Added References

It is possible that the articles submitted to the journal may be augmented by a textual analysis which will add hyperlinks to the document. It may be that the reference section may also be augmented through analysis to add pointers to online copies of documents referenced. Authors, Titles or Subjects could be checked via online collections of bibliographies.
This augmenting can also be seen as non-intelligent editorial interference so great care has to be taken. It is envisaged that copies of the "new" article be made available online to the author before it is published so that they may "sign it off".

Although SABS is still under development at the time of writing of this paper, we expect that the value of a Software Agent Based System will be demonstrable at paper presentation time.

The Future   

Success or failure will not lie with the mid-year issue but with the acceptance of Quality Multimedia based electronic journals per se. I think that the future has the rosy glow of electrons.....

References