Local time: Friday, 05-Dec-2008 12:24:39 EST
Last update: at /special/conference/apwww95 , Friday, 21-May-2004 09:47:09 EST

Exploiting the World-Wide Web for Electronic Meeting Document Analysis and Management

Gitesh K Raikundalia

CRC for Distributed Systems Technology,
Bond University,
Gold Coast, Queensland,
Australia, 4229.

gitesh@dstc.bond.edu.au
http://www.dstc.bond.edu.au:8008/staff/gitesh.html

Michael J Rees

CRC for Distributed Systems Technology,
Bond University,
Gold Coast, Queensland,
Australia, 4229.

rees@dstc.bond.edu.au
http://www.dstc.bond.edu.au:8008/staff/michael.html

[1]

Abstract
This paper describes a means of Web browser access to various meeting document information, such as minutes, tabled documents, reports and document indexes. Applications are being developed to take standard electronic meeting log files, postprocess them in a variety of ways, and generate a series of indexes and summary files. These files are formatted in HTML and exploit hyperlinks to the full in order to relate the different types of information.
Keywords
World-Wide Web, Computer-Supported Cooperative Work, electronic meetings, document management, document generation.

Introduction

For conventional meetings someone must take responsibility for preparing and distributing paper copies of all documents required for a meeting. Typical examples are agendas, minutes and supporting papers. Providing each meeting participant has access to a workstation running appropriate electronic meeting software during the meeting then the Web can be used to:

Note that there is still considerable effort involved in preparing meeting documents in HTML format.

This paper describes a mechanism for generating a number of the documents required for a meeting using a software tool for electronic meeting document management. To be effective this tool takes as input a verbatim record of all remarks (the actual conversation) typed during the meeting. Most electronic meeting systems generate such a meeting log file. LoganWeb, the electronic meeting document management tool described below is designed to generate a number of polymorphic documents by performing a series of analyses on the meeting log file as shown in Figure 1. Note that a number of different types of meeting log polymorphs can be made available which leads to a management problem for the meeting participants when they access these documents in subsequent meetings.


Figure 1 Generation of polymorphs

Algorithms for generating some of the more complex polymorphs are still under development. From previous preliminary work described in [RaA95] it is clear that only semi-automatic analysis will be possible. To be fully effective, it will necessary for one of the electronic meeting participants to act as a secretarius. The secretarius will be responsible for injecting into the meeting log a series of cues which relate the topics and agenda items under discussion with a series of typed remarks while the meeting is in progress. It will be possible for the meeting chairperson to act in this capacity for most meetings. During large, formal meetings, however, a separate secretarius will be needed.

During an electronic meeting it is envisaged that participants are in real-time communication from their workstations by running a tool such as the Yarn system described later. Each participant will be viewing two windows while the meeting progresses:

  1. the Yarn window showing the shared meeting remarks, the list of participants and mechanisms for submitting remarks, voting, and so on
  2. the Web pages generated by LoganWeb which allow participants convenient access to meeting documents via hyperlinks and interactive forms
The complexity of the range of meeting polymorphs, their relationships and the order of access during a meeting makes for a significant user interface design problem which will be discussed in more depth in a companion paper. The presentation and structure of the polymorphs, and the time of their generation to meet participants needs is described below. Early prototypes of LoganWeb have indicated that conciseness of display on a Web page and productive exploitation of hyperlinks play significant roles.

Electronic Meeting Phases

The context in which a meeting process takes place is a three-phase structure as shown in Figure 2. Such pre-meeting and post-meeting issues are discussed in [BeA94, CoA87]. Figure 2 covers the type of activity that would occur within the meeting phases when using Yarn and LoganWeb. This meeting structure would be just one example in a collection of meetings where documents generated from one meeting may be fed as input to another meeting. The use of LoganWeb is applicable in all phases of the structure, whereas the Yarn system would only be used within the actual meeting itself.


Figure 2 Three-phase meeting structure

The first phase is concerned with preparatory document viewing before the meeting. This would involve participants accessing and familiarising themselves with the meeting agenda and any documents of previous meetings appearing necessary for the meeting. During the meeting phase, the meeting system is used for conversation while any documents, such as the agenda, are referred to. The post-meeting phase consists of automatic generation of documents, such as verbatim minutes or concept categories. Documents such as summaries will be manually generated by the secretarius. A scenario exemplifying the real-world application of the meeting structure and associated activity is presented in [RaB95].

Meeting-Document Interaction Model

A simple, three-level model of interaction for describing a meeting process is proposed -- Participant, Meeting Software and Document Storage levels (Figure 3). The outermost level is concerned with the participants who are meeting together. The second level indicates the intermediary nature of software in accessing or generating documents for participants. This is the realm occupied by both Yarn and LoganWeb, and other software that can aid the meeting process, although it is firmly believed that the paradigm suggested is sufficient for virtually all meeting situations. The innermost layer relates to the method of storage of the documents.

The annulus structure of the model conveniently reflects the geographically distributed nature of meetings and the centrality of the site storing the documents. Participants such as those shown in Figure 3 may be located anywhere that Yarn and LoganWeb are accessible, ranging from all participants situated locally to all participants situated globally.


Figure 3 Meeting-document interaction model

The Yarn System

The electronic meeting tool used in this work is the Yarn meeting system ([ReA93, ReB93, ReA94, WoA94]). The system is conversationally-oriented where participants submit remarks that are read in a WYSIWIS ("What You See is What I See") fashion. The tool can be used either in synchronous or asynchronous modes.


Figure 4 Sample Yarn log

An example of a (partial) Yarn log is shown in Figure 4. Meetings are carried out within a channel, so that this log is specific for a particular channel. Remarks made by participants are contained within a pair of dashed lines with the participant's name inside the first of the pair and alongside the lines of the remark. Timestamps are also recorded throughout giving accurate information about the timing of the remark. These logs are the major documents associated with meetings and it is the log analysis that yields information in different forms useful to meeting practice.

Meetings are intended to follow a scheme of topics and further sub-topics by the use of recognised keywords. The start of a topic, for example "wscrawl experiment", is shown at the top of the log in Figure 4. A sub-topic within this topic could be commenced by simply starting another topic in the same way, carrying out the discussion, and concluding with an agreed end keyword.

LoganWeb

Web pages providing access to meeting documents are generated by LoganWeb ([RaB95]). This body of information consists of a home page (as shown in Figures 6 and 7) that is composed of sections giving access to document indexes, direct access to documents or the capability of immediate generation of documents. The first section provides an entry point to the indexes, which allow traversal of a handful of relevant pages so the user arrives at the desired document. Document generation is performed once by the secretarius resulting in the documents being stored at the central site.

In order to implement this functionality, the Web user interface mechanisms of image-based selection (of information), option menus, checkboxes and fill-out forms as well as the usual hypertext linking are employed. The last part of this paper will present user interface designs of LoganWeb explaining the use of this functionality.

The purpose behind LoganWeb extends beyond mere presentation facility as found usually with meeting information placed on the Web, such as at [ScA95, OrA5]. The browser acts as an interface to an automated document generator, implemented by Common Gateway Interface (CGI) scripts. These scripts are executed for Web page creation, handling the aspects of page layout, content and linking. Thus, the media that is used for presentation of documents is also the same media used for analysis and the manufacture of these documents.

Document generation involves the treatment of logs such as that shown above as input and produces auxiliary documents or the concept categories document as output (refer to Figure 1). For each polymorph there is a separate requirement regarding content and accordingly a separate method of "massaging" or altering original log content to satisfy this requirement. The method of organising polymorphic documents in storage is simple and intuitive. The documents are arranged according to the fields within the log name which is:

<channel>.<year>-<month>-<day><alphchar>
where channel is the meeting channel used to generate the log, the year, month and day are the date of the log and alphchar is an alphabetic character used to distinguish the log from multiple logs of the same channel and date [3]. An example of a log is BUdstc.94-Feb-22b which is the second log produced on the 22nd of February, 1994 for the channel BUdstc. The day directory is where the documents are actually stored. A directory structure associated with a file system has been adopted as the document repository. Reasons for this are: As a result, an alternative database solution has not been applied.

The interaction between LoganWeb and the directory structure storing polymorphs is shown in Figure 5. One action is the access of the polymorphs for display to the participant. This clearly does not alter the directory structure in any way. The other action is polymorph generation, when the document is placed in the appropriate lowest level (day) directory of the hierarchy.


Figure 5 LoganWeb-storage structure interaction

Document Types

There are two types of documents proposed, polymorphs and other documents. Polymorphic documents include: The other documents are: Other miscellaneous forms of documentation should be accessible through LoganWeb such as reports, listings of program code or scanned journal articles. Because there is no limit to the type of meeting or what may be discussed during the meeting, anything that may be documented and assists participants should be supported via LoganWeb. For this access to be achieved, the document will need to be presented via HTML [2], or at least use the default capability of the Web whereby a document not marked up in HTML may be presented in the browser as plaintext (where Web facilities such as hyperlinking are obviously not available).

LoganWeb Page Generation

Figures 6 and 7 show the first two and last two sections of the four-section home page, respectively. The first section, "Meeting document indexes", is composed of a listing of years. Clicking on one of these years takes the participant to a Web page with a set of calendar months for that year, such as that shown in Figure 8. Calendar-driven document selection is manipulated because it is a standard technique for selecting information associated with a certain date. Humans are familiar with the use of calendars in everyday life, providing a convenient real-world metaphor for this application. The image map capability links a date of the calendar with the meeting document index for that particular date.


Figure 6 LoganWeb Home page - first two sections


Figure 7 LoganWeb Home page - last two sections

An index such as that in Figure 9 is shown to the participant after clicking on the desired date. For each meeting channel used on this day there is a grouping of documents for each log: polymorphs generated from the logs and other applicable documents. If a particular document does not exist, it will not be displayed in the index listing (for example, a particular polymorph that has not been generated from the log). Clicking on one of the document names will bring the participant to the end of the Web page traversal for index-directed document access -- the actual document selected will be displayed.


Figure 8 Calendar index to 1995 documents

The document, in this case the verbatim minutes, would appear as in Figure 10. At the bottom of the document will be a list of links to other documents of the same type in existence; in this case, a polymorph is being viewed so links to other polymorphs already generated are given.


Figure 9 Document index of 5 June, 1995


Figure 10 Verbatim minutes of 5 June, 1995

Direct Document Access

In referring back to Figure 6, the second section is the one concerned with providing direct viewing of documents. That is, the aforementioned indexes can be bypassed with immediate access to the desired document. The Web interface method used is that of the fill-in form based on option menus. Here a participant may set the properties associated with the document and then submit these values. Menus exist for selection of meeting channel and date. Because user input is required for meeting section and keyword polymorph information, forms and checkboxes are used for selecting the desired document. Figure 6 shows the example of a participant selecting the verbatim minutes polymorph for the "tools" meeting of the 5th June in 1995. This would lead to direct access of the document shown in Figure 10.

Document Generation

The secretarius is the participant delegated the task of generating polymorphs to be viewed by any group member. The third section of Document Generation, as shown in Figure 7 (where the browser window size has been resized as smaller to ease understanding of the interface) is used by the secretarius for this task where multiple documents of the current meeting may be created in one step without having to actually view any document immediately. This means that Web page traversals, which are unnecessary, are avoided and the outcome is simply the existence of new documents. Generation of a new document also occurs through the Direct Document Access section if the document does not already exist, but this is a more cumbersome process if there are many polymorphs that need to be created. As in the second section, forms and checkbox interfaces are exploited for the same reason.

Document Linking

It was demonstrated above by referring to Figure 10 how links can connect documents of the same type together. Figure 11 shows a keyword polymorph which illustrates the linking between: For reasons of space, such linking for all polymorphs will not be shown.


Figure 11 Keyword polymorph for "scan"

Figure 11 is generated as a result of searching a log for all occurrences of the string "scan". The hyperlinked headings, such as "[Topic] Web page advertisement", are composed of two parts -- the:

For presentation purposes the entire heading collects together the set of keyword search results for the topic listed under it. Functionally, the "[Topic]" part provides a link to the start of the topic in the corresponding topic polymorph, and the rest of the heading links the participant to the start of the topic in the corresponding log.

Timestamps are shown underlined and above a line of a remark and connect the participant to the start of the remark in the log where this timestamp is positioned. Finally, each occurrence of the keyword in the polymorph provides hyperlinked access to the line in the log where the keyword occurs. Hence, the keyword polymorph provides the single line of the remark used in the meeting, and the link takes the participant back to the actual context of discussion, providing him/her with more information if the single line is not sufficient for complete understanding.

Conclusion

The Web browser interface is highly effective in the management and presentation of documents in electronic meetings. The user interface of a Web browser can be manipulated very conveniently during a meeting process, in any of the three phases, to view documents, and move from one to another as necessary.

Browsers support a range of interface mechanisms, such as hypertext and image-based selection and fill-out forms, that are ideal for access to and generation of documents. Examples of user interface designs for LoganWeb pages have demonstrated Web hypertextual capability being exploited for linking related documents. The nature of hypertext is conducive to index structuring and functionality and has been used in a three-level index to documents. Such capability may also help the user arrive at the exact information within documents that he or she desires in a very short time. Certain textual parts of polymorphic documents, ordinarily having a simple meaning within the document, may be assigned special linking roles (for example, timestamps in a keyword polymorph); these roles connect sections of the polymorph to the log context from which they are derived.

Exploitation of the Web interface relies on treating the browser as more than just a document viewer but also as a document generator by using CGI scripts. Such generation inputs a raw Yarn log and uses an appropriate algorithm to derive the desired polymorphic document. Further work is being undertaken in the creation and implementation of these scripts.

Footnotes

[1]
The work reported in this paper has been funded in part by the Cooperative Research Centres Program through the Department of the Prime Minister and the Cabinet of Australia.
[2]
Converters exist for converting documents in several forms, such as LaTeX or Rich Text Format (RTF), into HTML although the results of such conversion may not be quite satisfactory.
[3]
The ordinal value of this character indicates the chronological order of the generation of the log from all logs of that day.

References

[BeA94]
Bergmann, N. W. and Mudge, J. C., "Automated Assistance for the Telemeeting Lifecycle", ACM 1994 Conference on Computer Supported Cooperative Work, Chapel Hill, 22-26 October, 1994, 373-384.
[CoA87]
Cook, P., Ellis, C., Graf, M., Rein, G. and Smith, T., "Project Nick: Meetings Augmentation and Analysis", ACM Transactions on Office Information Systems, Vol. 5, No. 2, April 1987, 132-146.
[HuA95]
Hughes, K., "SWISH documentation", World Wide Web, http://www.eit.com/software/swish/swish.html.
[OrA95]
"Organization of American States 1995 General Assembly - Annotated Agenda", World Wide Web, http://www.oas.org/ga3.htm.
[RaA95]
Raikundalia, G. K. and Rees, M. J., "Preliminary Investigation of Concept Categorisation from Electronic Meeting Logs", CRC for Distributed Systems Technology 1995 Symposium, University of Queensland, 6-7 July, 1995.
[RaB95]
Raikundalia, G. K. and Rees, M. J., "A Scenario of a Web User Interface Tool for Electronic Meeting Document Generation and Presentation", QCHI95 Symposium, Bond University, 21 August, 1995, http://coral.dstc.bond.edu.au:8008/general/qchi95/, to appear.
[ReA93]
Rees, M. J., Smith, G. P., Iannella R., Lee, A., and Woo, T. K., "Yarn: Text-based Electronic Meeting Tools in a Distributed Environment", Computational Support for Distributed Collaborative Design, University of Sydney, 17 Sept. 1993, 3-20.
[ReB93]
Rees, M. J., Iannella, R., Lee A., Smith G. P., & Woo, T. K., "Spinning a Yarn: User Interfaces for Synchronous Remote Electronic Meetings", Proceedings of OZCHI 93, University of Canberra, 25-26 Nov. 1993, 42-58.
[ReA94]
Rees, M. J. and Woo, T. K., "A World-Wide Web User Interface for an Electronic Meeting Tool", Proceedings of OZCHI 94, Melbourne, Australia, 1994, 187-192, http://www.dstc.Bond.edu.au:8008/research/reports/html/OZCHI94/ozchi94s.html.
[ScA95]
"Scientific Steering Committee Minutes", World Wide Web, http://www.usglobec.berkeley.edu/usglobec/minutes/minutes.home.html.
[WoA94]
Woo, T. K. and Rees, M. J., "A Synchronous Collaboration Tool for the World-Wide Web", Proceedings of the Second International World-Wide Web Conference, Chicago USA, 17-21 Oct., 1994, http://141.142.3.70/SDG/IT94/Proceedings/CSCW/rees/SynColTol.html.
[Return to Table of Contents]
COPYRIGHT © 1995 by AUUG95 and APWWW95 Charles Sturt University. ALL RIGHTS RESERVED. ISBN 1 875781 43 9