Reprint: Electronic Journals, 1983

7th International Online Information Meeting, pp 111-.118,  1983

A SOFTWARE PACKAGE FOR
ELECTRONIC JOURNALS

C.F. Reynolds, BruneI University, UK

Keywords: Interactive Papers, Electronic Journals, Computer Aided Learning, Authoring Systems, User Friendly Systems.

Abstract: In connection with the British Library funded electronic journal, "Computer Human Factors", a fully  interactive paper was written involving a number of experimental features. The "CHF" paper describes a teaching package used to introduce undergraduates to computers.Readers of the paper are provided with a number of general tools - for instance they can locate texts containing given words or write on-line notes for their own use, or to send to the author. In addition the paper contains a number of appendices in which a high degree of interaction is possible. For instance the reader can actively run the student coursework with whatever input he chooses. Interactive tutorial texts  and support statistics are available (in some cases the figures are produced by searching a data base online). The reference list allows the reader access to abstracts or short commentaries where this would be helpful - and for one key reference the whole paper has been represented in interactive form as an appendix.

1. INTRODUCTION

This paper describes the principle features of a software package lIritten to support a fully interactive paper, "An interactive Data Base System Designed to give Terminal Confidence" (Ref. 1) which was written in conjunction with the electronic journal "Computer Human Factors", although it uses different software. The first section discusses the background to the British Library Project, and the software used to support the paper. The design factors leading to the fully interactive paper are given, followed by a description of the principle features of the electronic text. Finally the principle findings of the research are given

1.1 LINC and BLEND

In 1980 the British Library Research and Development Division decided to fund a major project to investigate the potential of fully interactive publications, in which the authors wrote their documents online, the journal editor and the referees read them, and the text was finally published for online journal readers. A detailed description of this project has been published by the chief investigator, Professor B. Shackel, of Loughborough University (Ref. 2). The author is a member of the community of scientists associated with the "Computer Human Factors" electronic journal. This community is called LINC, short for Loughborough Information Network Community.

The central computer facilities associated with this network are based on the DEC 20 computer at Birmingham University. The combination of hardware and software is called BLEND, or Birmingham and Loughborough Electronic Network Development. Some LINC members, including the author, have British Library funded Torch computers to provide data communication and local word processing facilities.

1.2 NOTEPAD

NOTEPAD is a commercially available conferencing package produced by Informedia Corporation and provides the software support framework for the BLEND system. The package is limited to the exchange of paragraph sized chunks of text within a simple project/activity structure, and it lacks many of the features of conferencing systems such as EIES (Ref. 3). Two limitations significantly affected its use in the LINC/BLEND environment. Its inbuilt editing facilities were barely adequate for the kind of ephemeral notes which people exchange in conferencing situations and hence totally inappropriate for the preparation of texts of publication standard. This leads to most authors preparing papers outside NOTEPAD, often using conventional paper writing methods, and never seeing their manuscript in interactive form until it had been completed and transferred to NOTEPAD.

The other problem was the division of papers into strictly sequential numbered paragraphs, which, coupled with the reading facilities within NOTEPAD, restricted the logical structure of the resulting text. This has been partially overcome by the writing of an external package, by T. Maude, of Birmingham, which allows papers to be read in a more versatile manner, although it still considers the paper to be a series of sequentially numbered paragraphs.

Once these limitations in NOTE PAD were realized there was a vigorous exchange of correspondence among the active members of the LINC community as to the features one would like to see in a "next generation" electronic journal. It soon became apparent to the author of this paper that many of the "desirable" features were already available on software being used, for other purposes, at Brunel University. In addition several other requirements could easily be met.

1.3 CODIL

CODIL is an information processing system specifically designed for mathematically unsophisticated users with poorly structured and open ended tasks. Details of the prototype software have been published (Refs 4, 5) while an up to date account of the current operational package is in preparation. The philosophy behind the approach is given in Ref. 6, while Ref. 7 gives a summary of some of its many applications.

1.4 The Teaching Data Bases

An early practical use of CODIL was to support a data base containing interactive lessons on how to use the language (Ref. 8). As soon as fully interactive facilities became generally available for teaching at Brunel the language was used to set up a series of teaching data bases. To do this a computer aided lesson generator was constructed, which made it easy to set up blocks of text linked by menus, simple questions and answers, user written routines, etc. This has been used to support an introductory questionnaire of students' prior computing experience, data bases containing information on reading lists, department staff And students, and interactive lessons on COBOL The software described in the bulk of this paper is an upward compatible version of this teaching package.

1.5 FIXIT

FIXIT is a CODIL data base used to introduce undergraduates to the power of a computer. It is loosely based on a package written for a British television program called "Jim'll Fix It" where it was used to show a 12 year old how a computer could help him to do his homework. The undergraduate coursework, which is to assess the suitability of FIXIT for a U year old, is the subject of the interactive paper described below (Ref.1) and is also to be described in conventional "paper" format (Ref. 9).

The package consists of a number of separate activities called Bird, Calculate, Crisis, French, German, Kings, Nim, OXOXOX, Quiz and Riddle Me Ree. These are 'linked into a moderately user friendly framework. Each activity is designed to demonstrate some aspect of computer science (although the students are not told this until after they have done the coursework) and some deliberately user awkward features which the students are meant to identify.

2. DESIGN CONSIDERATIONS

In considering the design of software to support fully interactive publications it is important to start with printed papers, if only because initially authors and readers will be familiar with the layout. This suggests that each paper should have clearly identified "title, abstract, introduction and conclusions. There will be the "body" of the paper, and possibly tables, figures, appendices and references depending on the subject. There should be an advised route through the body of the text for those who want to study the paper exhaustively, together with aids to selecting sections of the paper of particular interest. The latter can be considered to be the interactive equivalent of browsing through a printed paper.

On top of this there is a considerable potential for introducing new types of interaction. There is a considerable opportunity for expanding the concepts of references and glossaries. When a reader comes across a citation he should obviously be able to get the full bibliographic details. However, the option could be given to see either an abstract or, if the cited paper is also fully interactive, relevant parts of the paper itself. Families of papers on selected subjects could share bibliographies and glossaries.

Often a scientific paper represents no more than an analytical summary of a significant quantity of observations. Conventional printing techniques make it impossible to include such bulky data but in many cases it could be held as an online data base, allowing the reader to extract and process the information he considers to be the most relevant to his needs.

One apparent limitation of the online paper is the problem of making marginal notes. Clearly the software should keep a record of the parts of the paper you have, or have not seen, allow you to make readily retrievable notes, and even to issue "public" notes, for other readers.

For papers which describe computer software there is another possibility, which is a key feature of the paper written for "Computer Human Factors". This is the ability to include the software as a dynamic part of the paper. This means that the reader is not constrained to the simple printed examples of a conventional publication as he can "run" his own.'

3. THE "PAPER" PACKAGE

The software package used to set up the experimental paper for "Computer Human Factors" started as a computer aided learning data base generator, and is still used for that purpose. It is written in CODIL and consists of a series of internal support routines, a default set of HELP and other user support facilities, and a series of authoring routines. Under normal CAL use each teaching package is set up with a separate data base of linked topics. In the experimental paper a number of topic data bases have been set up as appendices, and facilities provided which allow readers of the paper access to the student teaching data bases, without the students being able to move in the opposite direction

3.1 Authoring Facilities

The author can extend the lessons/paper by typing "NEW TOPIC". He is then asked for the name of the topic, whether the name should appear in the initial reader's index, and the heading to be printed out, if any. The author can then type in a series of commands, the ones most relevant to authoring being given in Table 1.

Command		Action
ask		Asks the reader to type in a response.
change		Changes the reader's data base
direct		Direct insertion of CODIL code into the topic.
do		Calls for the named topic to be executed.
end		Terminates the topic or current response block.
file		Inserts "subroutine" call to CODIL file
line		Allows text to be input as independent lines.
menu		Defines a series of options and their responses.
new page		Author controlled paging of the reader's VDU.
offer		Offer a named topic to the reader.
remember		Add topic name to the reader's active topic list.
response		Defines an answer to the previous question and sets up a response block for the relevant action.
text		Allows the input of continuous text
yes or no		Accepts only "yes" or "no" responses.

    Table 1. Some Authoring Commands.

The author can arrange for the reader to be given free choice of topics, with the aid of a topic list, or to be started in a given topic, and transferred to others by his response to questions, probably using menus. The former is more appropriate in the interactive paper environment while the latter is best for comparatively closed teaching tasks. It is also possible for him to define synonyms to catch the reader's more likely "mistakes".

While new texts, etc., can be added by the author in a fully interactive mode updating these texts is not as easy as it might be. This is because the CAL package was designed to support a number of simultaneous users and the topics are "compiled" to give run time efficiency. This makes editing difficult and source texts are normally kept to regenerate the paper. Thi is not a necessary restriction but it was decided that effort shou1d be concentrated on the reader aspects of the research.

3.2 Reader "HELP" facilities.

When the reader has free choice of topics he will normally have access to a series of help facilities. These will guide him through the features described later, give him information about the best use of his terminal, and refer him to any additional help topics provided by the author.

3.3 Finding One's. Way through the Paper

The reader of an interactive paper needs an active guide to help him to find his way. The author of the paper can make it easier by providing suitable contents list, and the contents list for the main paper is given in Table 2. At any stage he can type "list" to get a list of all the topics in the current data base known to him. Initially these will be the ones the author said should be indexed, and the list should be comparatively short. As the reader explores the paper other topics, such as citations, will be introduced and automatically added to the list. At any time the user may check which topics he has "seen" and which still remain unseen.
where the paper is divided into separate data bases/appendices separate lists are kept for each, as a combined list could become impossibly long.

CONTENTS
abstract
contents
section 1		Introduction
section 2		The BruneI University Computer Science Course
section 3		The University Computing Facilities
section 4		The Students on the Course
section 5		The First Year Computer Science Module
section 6		Introducing the Students to the Computer
section 7		The Background to FIXIT
section 8		The FIXIT Package and Coursework
section 9		Observations
section 10		Conclusions
appendices
acknowledgements
author

Table 2. The Contents list of the Experimental Paper

An additional facility, "locate", generates a list of topics which contain a user supplied word or phrase.

3.4 Changing Direction from a Visual Display Unit

Normally the user of a visual display unit will ask for the text of the paper to be passed to him in convenient sized pages. The package contains its own facilities which allow the user to start a new screen, scroll or quit. In addition the user may request additional topics, returning to the branch point later. This can be a useful way of following up references and glossary entries. The user may also use this opportunity to recall a table, etc., he has seen earlier.

3.5 User Note Taking Facilities

The online paper reader clearly wants to keep any notes he makes with the paper. The current software allows the user to flag a topic and, perhaps more usefully, allows him to make a series of notes. Both are kept from run to run so that they are automatically available to the user at a later date. These notes are private to the user. An index is kept of these notes, and additions, amendments and deletions can be made at any time. The facility has proved very useful to the author as flaws found during the interactive equivalent of proof reading can be recorded online as they are spotted, and deleted as soon as they have been corrected. Journal referees could use the note facility in a similar manner.

In addition the reader can send comments to the author. These may then be added to an appendix dedicated to reader comments. In theory it would seem appropriate to allow readers freedom to make such public comments directly, but in a refereed environment some constraints are needed.

3.6 Incorporating Extant Software

One of the appendices of the paper contains the FIXIT software and the reader may transfer into it and run any activity as if they were a student doing the coursework. At the same time students may be running the FIXIT coursework, using the identical files, but they cannot transfer to the PAPER. To give the reader greater flexibility individual FIXIT activities may be entered from the body of the interactive paper, so that he may follow up details of the activity he has just been reading about. This degree of dynamic flexibility is almost impossible to illustrate effectively in a short printed paper, simply because the whole essence of the activity is the spur of the moment decision of the reader to follow up a side issue.

3.7 Information

One appendix contains detailed discussions of each of the FIXIT activities, including access to "improved" versions which do not include the deliberately user-unfriendly features. In addition it contains a number of examples on information retrieval from files related to the paper. In some cases, for instance the ability to examine the riddles in Riddle Me Ree, the trivial content is trivial and it is simply included to show the potential of the approach. Elsewhere the statistics of the students on the course is dynamically evaluated when requested - but a "frozen" copy of the file is now used as the paper's statistics changed every time a student withdrew from the course:

3.8 References, the Glossary, and related papers

The principal topics in the appendices REFERENCE and GLOSSARY have been made global in scope - so that a reader of the paper can, for instance, get a definition of BLEND whenever he wants.

Many of the references have abstracts with them, which may be displayed if the reader requires. In addition one full and one short paper have been incorporated in full, with minor modifications to facilitate interactive reading. These share the REFERENCE and GLOSSARY with the main paper.

4. OBSERVATIONS AND CONCLUSIONS

At the time of writing the interactive paper has only been tested on the BruneI University computer, although it is planned to transfer it to Birmingham and integrate it with the BLEND system very soon. Observation are therefore confined to student testing and LINC members who accessed it Brunel computer directly.

Almost all users found the system extremely user friendly and the few who had problems seem to have "got lost". The reason for this seemed to associated with the initial help text, and highlights a difficulty with prototype fully interactive papers. The trouble is that many of the readers will not be particularly interested in the subject matter of the paper - they simply want to tryout the interactive facilities: The introductory text was aimed at people with an interest in introducing undergraduates to interactive computing and suggested that the reader might like to do the student coursework "blind", and read the paper after wards. This was a mistake and it is clear that the introduction needs to allow for the different needs of:

(A) The novice reader interested in the subject of the paper.

(B) The interested reader with prior experience of interactive texts who wants to take notes, send messages to the author, etc ..

(C) The reader who is disinterested in the content of the paper, but who wants to explore the potential of interactive papers in general.

Several readers have commented on the constraints of the VDU screens, and how this affected the way the paper was written. They agreed with the approach the author had taken on the problem of style. It would obviously be pedantic in the extreme for the system to ask "What does the reader want to do next?". However, once the reader is called "you" the conventional impersonal approach to paper writing becomes inappropriate. The constraint of 20 lines of 80 characters of usable space per display, coupled with the ability to approach topics from different directions, also affects style. This is not always a disadvantage. For instance the definition of terms, such as BLEND, can be put in the GLOSSARY. Readers who know about it need never see this text, while the others can see a longer definition than could be given in a conventional paper at any time they want. Footnotes, and discussions about the principle references, can be treated in a similar way.

The ability to interact directly with the software being described is a major advantage to the reader. Often, on reading a paper describing some software one is uncertain whether the simple example was chosen because it was short enough to be acceptable to the journal editor, or because it represents all that actually worked at the time the author decided to publish. When an interactive paper contains the software it describes, the reader can try it out for himself on examples relevant to his own interests. Similar factors apply if the paper contains a data base of raw observations which the reader can analyse himself. The ability of the reader to communicate with the author directly is also an advantage. From the point of view of the second rate author these "advantages" may be seen as disadvantages. His software will need to be far more robust than is usual for most research packages, and any data that he makes available for online interrogation, etc., must be of high quality. Under such circumstances it seems likely that only the more competent authors would want full use of interactive papers of the kind described above.

To conclude, it seems very likely that fully interactive journals, in which the authors made their software and raw data available to the reader, could have considerable advantages to the reader. They require a far more vigorous approach by the author who may well be tempted to take the easier route of the more conventional publication techniques. Because of the extra effort an author has to put in to writing such a paper he is unlikely to do it unless the interactive journal is widely available to those who would find it of interest.

ACKNOWLEDGEMENTS

While the interactive paper described in this paper was developed using Brunel University facilities, the inspiration is due to the
author's membership of the LINC community. Thanks is therefore due to British Library, who funded the project, the project teams at Loughborough and Birmingham, and my colleagues in the LINC community. It would not have been possible without the many students who did the FIXIT coursework, and the smaller number who test read the interactive draft.

REFERENCES

1. Reynolds, C.F. "An Interactive Data Base System Designed to give Terminal Confidence", Computer Human Factors, (in press).
2. Shackel, B. "The BLEND System - Programme for the Study of some Electronic Journals", Comp. J., Vol 25, 161-8, 1982.
3. Hiltz, S.R. & Turoff, M., "The Evolution of User Behaviour in a Computerised Conferencing System", CACM, Vol 24, 739-51, 1981.
4. Reynolds, C.F., "CODIL, Part 1, The Importance of Flexibility", Comp. J., Vol 14, 217-20, 1971.
5. Reynolds, C.F., "CODIL, Part 2, The CODIL Language and its Interpreter", Comp. J., Vol 14, 327-32, 1971.
6. Reynolds, C.F., "A Psychological Approach to Language Design", Proc. Workshop Computing Skills and Adaptive Systems, Liverpool, 77-87, 1978.
7. Reynolds, C.F., "CODIL as an Information Processing Language for University Use", IUCC Bulletin, Vol 3, 56-9, 1981.
8. Reynolds, C.F., "Teach Yourself CODIL", BruneI University, 1976. (An updated interactive version of these interactive lessons are available with the CODIL software.)
9. Reynolds, C.F., The FIXIT teaching package, (in preparation).