DISCOVERY OF MEETING-PARTICLE LINKS AND THEIR

APPLICATION TO MEETING RECOLLECTION SUPPORT

Ishitoya Kentaro

, Ohira Shigeki

and Nagao Katashi

Graduate School of Information Science, Nagoya University, Nagoya, Japan

Information Technology Center, Nagoya University, Nagoya, Japan

Keywords:

Context discovery, Meeting particle link, Periodic meeting, Meeting recollection support.

Abstract:

To facilitate more efﬁcient regularly-held meetings, it is important to consider the past discussions and recall

them in the current discussion. We previously developed a meeting recording system that creates discussion

content from casual meetings on the basis of digital whiteboards. In this paper, we describe a discussion tool

that enables the content of past discussions to be structured, retrieved, and reused in future meetings. We

developed an application system to visualize the context relating different meetings by discovering links be-

tween meeting particles that are fragments of meeting content (e.g., text, image, and sketch. This visualization

enables participants to recollect past discussions in the current discussion.

1 INTRODUCTION

Figure 1: Meeting with TimeMachineBoard (Using a pen-

device to point at context).

Organizations like companies and universities hold

many meetings, large and small. People participate in

meetings to solve problems they face or to get good

ideas from other participants. They carry out their

work on the basis of what was discussed in such meet-

ings.

Small periodical meetings whose members have

a common goal or purpose, such as team or project

meetings, are especially important. Making these

meetings better will lead to better company perfor-

mance and research.

Meetings, as a tool for organizations to get things

done, have certain problems. One problem is when a

meeting goes on for several hours; it is difﬁcult to cre-

ate the minute, and as a result, the content may easily

be forgotten. This situation is made worse when the

relationships between previous meetings and the cur-

rent meeting are not clear; we can easily lose focus

of the topics, and this can lead to a repetition of the

same arguments. Many readers are no doubt famil-

iar with meetings where someone brings up the same

topic that had seemingly been settled at a previous

meeting. Even if we had all the minutes of the past

meetings, such a problem entails lost time. To solve

it, we must consider the relationships between meet-

ings.

We have been researching the contexts of periodi-

cally held meetings. We call such meetings “casual

meetings” and are developing technologies to sup-

port them. We have already developed a casual meet-

ing support system called TimeMachineBoard (Ishi-

toya et al., 2009). This system enables participants to

show their drawings, images, texts on a large display

and record the discussion. The recorded discussion

is composed by these elements and their arrangement

histories, video, and audio. Figure 1 shows the ap-

pearance of our system being used in a meeting.

Our system aims to enable participants to retrieve

past discussions that would be relevant to the current

discussion. Participants can then base their discus-

456

Kentaro I., Shigeki O. and Katashi N..

DISCOVERY OF MEETING-PARTICLE LINKS AND THEIR APPLICATION TO MEETING RECOLLECTION SUPPORT.

DOI: 10.5220/0003659904480451

In Proceedings of the International Conference on Knowledge Discovery and Information Retrieval (KDIR-2011), pages 448-451

ISBN: 978-989-8425-79-9

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

sion on the past content as need be, and they can view

recorded content and be aware of the relationships be-

tween the past and present meetings.

We developed a tool to enable users to bring all

content of past meetings into the current one, search

the content for certain topics, and bring them into the

current discussion. The tool extracts contextual in-

formation from the operation by discovering links be-

tween fragments of past meetings, e.g., texts, images,

and sketches, that we call meeting particles. We also

developed an interface for visualizing this contextual

information. This interface enables participants to

check the content of past meetings during a discus-

sion and make their meetings more productive.

2 RELATED WORKS

Research on small informal meetings has often con-

centrated on how people use electronic whiteboards,

such as Liveboard (Moran and van Melle, 2000;

Oguni and Nakagawa, 1996; Streitz et al., 1994).

These studiesf electronic whiteboards were projec-

tors or large back-lit displays. Their user interfaces

were digital pens and boards from which they derived

semantics from the strokes and texts written on the

board.

Research has also been done on recording white-

board content. DYNOMITE (Wilcox et al., 1997)

records content and related sound and information.

ReBoard (Golovchinsky et al., 2009) monitors the

whiteboard with a camera and if it detects that the

whiteboard content has been modiﬁed, it creates a

snapshot image of the whiteboard. The authors of

that study built a snapshot browser to share and search

white board content.

Other research has focused on regularly-held

meetings. Branham et al. considered how to reuse

whiteboard records after meetings and what values

whiteboard content has. They concluded that white-

boards act an information repository (Branham et al.,

2010).

3 MEETING RECORDING

SYSTEM

We focused on meetings that were regularly-held by a

university laboratory research project, and we specif-

ically wanted to ﬁnd ways to record, accumulate, and

reuse discussion content.

Our system, shown in Figure 1, uses a large dis-

play (hereafter, board) as a whiteboard to present the

content of a discussion for participants to share.

The board facilitates the present discussion by par-

ticipants sharing the content in the form of handwrit-

ten characters, ﬁgures, text, and images displayed on

a board (hereafter called, “display elements”). Partic-

ipants can variously arrange these display elements as

the meeting progresses.

Figure 2: System overview.

Figure 2 outlines the system. To display infor-

mation on the board, each participant can use a pen,

a pointer, or our developed software called iSticky.

Participants can write words or ﬁgures and move and

scale display elements directly by using a pen. They

can use a pointer to point to, scale, or move display

elements while they are away from the board. iSticky

is installed on the pad devices. Participants can use

iSticky to display, scale, and move display elements

(images, sketches, and texts) on the board. Each

pen, pointer, and iSticky has a unique ID. Thus, we

can ﬁnd out which participants introduced or operated

with which display elements.

Recorded discussion content is made up of display

elements and their histories of when and who input or

operated which display element on which board. At

the end of the meeting, these information is sent to

the database server and stored as searchable content

(called discussion content). By using this system in

meetings, we can easily record discussions.

TimeMachineBoard has been in operation since

September 2008. Four research projects have been

conducted at our laboratory (20 to 30 meetings per

project per year). I have made improvements the

system in response to various feedback that came

from these meetings. We can browse and search 157

project meetings held from September 2009 to Febru-

ary 2011. The average meeting time was 2 hours 12

minutes. Information displayed on the board con-

sisted of 244 strokes, 3627 text display elements, and

DISCOVERY OF MEETING-PARTICLE LINKS AND THEIR APPLICATION TO MEETING RECOLLECTION

SUPPORT

457

445 image display elements. Text display elements

transferred in a single meeting averaged 23 display el-

ements and 60 characters in Japanese (8 to 10 words

in English) per text display element. These results in-

dicate that TimeMachineBoard displays much more

texts than a whiteboard can.

4 DISCOVERY OF MEETING

PARTICLE LINKS

Figure 3: Interface for search meeting contents.

A discussion using TimeMachineBoard is different

from a discussion with a whiteboard. This system

is premised on discussions being recorded and later

searched and read. The recorded discussion is easier

to review than a record of discussion on a whiteboard

such as snapshot image of the whiteboard. This is be-

cause participants often use textual display elements

to conform their vocabulary to everyone else’s; this

makes the content easy to search.

We thought that subsequent meetings could be

made more productive when each participant can

effectively utilize solutions and knowledge derived

from the prior meeting.

For this purpose, it is necessary to enable dis-

cussion content of past meetings to be searched, re-

trieved, viewed, and discussed.

To refer to past discussion content without the cur-

rent discussion stagnating, the past discussion content

should be able to be searched and viewed quickly.

With this in mind, we developed the search interface

shown in Figure 3 as a function of iSticky.

The top of the interface is a board image view that

chronologically arranges the board images which ex-

press the conclusions of past discussions. The list of

the discussion contents can be perused by scrolling

right and left, and the board image expands when se-

lected. The list can be narrowed down by searching

using keywords, project information, participant’s in-

formation, etc. Moreover,discussion content of meet-

ings in which the users did not participate can be

downloaded from a server later and added to the board

image view.

Moreover, we thought that not only searching

and perusing past meeting content but also visualized

parts of the content (we call them “meeting particles”)

on the board in subsequent meetings would make par-

ticipants conscious of the context of the past meetings

and preventtopics being overlooked or the same argu-

ment repeated.

Meeting particles are compositional elements of

meeting content. Meeting-particles include not only

display elements themselves, but a part of a display el-

ement. For example, words in an image, morphemes

in a text, and speech in a recorded video. This en-

ables relationships between meeting contents to be

expressed more ﬂexibly and in more detail.

Therefore, to input to the board a meeting-particle

of past discussion content (we call this action “quota-

tion”), we created an element list view at the bottom

of the board image view that displays a list of display

elements (e.g., images, texts, etc.) contained in the

discussion content and selected by a user.

When a user selects the display element he/she

wants to quote from the element list view, informa-

tion about the selected display element is displayed in

a popup window. He/she can choose to quote directly

or to edit part of the content before they are input to

the board. For text display elements, morphemes in-

cluded in the input text are treated as meeting parti-

cles.

If a user quotes a meeting particle, it appears on

the board and the quotation information is recorded.

The linkage between meeting particles (information

regarding the relationship between the meetings) in-

cluded in the quotation information can be treated as

context associating the meetings.

Quotations using this interface were performed

158 times from September 2010 to February 2011.

Meetings held directly before the current ones were

quoted 108 times, those directly before them 39 times,

and older meetings 11 times. These results show that

the most recent meetings were most often quoted.

This means that useful discussions may get buried

over time as more meetings are held. Thus, a mech-

anism is required by which a participants can better

discover past discussion content.

KDIR 2011 - International Conference on Knowledge Discovery and Information Retrieval

458

5 MEETING CONTEXT

RECOLLECTION SUPPORT

5HODWLRQVKLSLQIRUPDWLRQ

GLVFRYHUHGIURPTXRWDWLRQ

Figure 4: Interface for visualization of meeting context in-

formation (blue lines indicate context information).

The search interface described above is for perusing

the content of one meeting, but it is difﬁcult for partic-

ipants to be conscious of the relationship between two

or more meetings. For example, by remembering past

discussions, when a participant can ﬁnd out a back-

ground of an implementation which is reported in the

current discussion, it would help to avoid unnecessary

repetition of the same arguments and opinions.

By searching past arguments and quoting a meet-

ing particle, information relevant to the past and

present meetings can be extracted from the quotation

information. To make participants conscious of the

context between meetings, we developed an interface

to visualize relationship information as a function of

iSticky.

The interface shown in Figure 4 displays content

in the form of nodes and relationship information in

the form of edges connecting nodes. This visualiza-

tion enables users to grasp how a series of discussions

has progressed from the past into the present.

Furthermore, users can quote meeting particles vi-

sualized on the interface. In comparison with the

board-image interface described earlier, this interface

enables more content from past discussions to be re-

discovered. Participants can quote by choosing a

node, since they are more conscious of the relation-

ships between meetings.

By using this interface during the meeting, as

well as retrieving the content of past discussions, par-

ticipants can become more aware of the context of

past meetings. Parts of useful discussion content re-

discovered by users can be displayed on the board to

be shared among participants, thereby making meet-

ings more effective.

6 CONCLUSIONS

We described a system for searching and quoting the

content of past discussions in subsequent meetings

and a system that supports retrieval of past discussion

contents by visualizing the context relating meetings.

Many researchers have analyzed the structure of

meetings, but few have concentrated on the linkages

between regularly-held small-scale meetings in the

knowledge-based activities of individuals and orga-

nizations. If relevant information from past meeting

records can be accumulated and used appropriately in

such meetings, any new discussion based on the past

discussions will make individuals and organizations

more productive.

Our future work includes evaluation of this sys-

tem through long-term use, automatic presentation of

information in accordance with the subject of a meet-

ing, and discovery of content of important meetings

by reusing information.

REFERENCES

Branham, S., Golovchinsky, G., Carter, S., and Biehl, J. T.

(2010). Let’s go from the whiteboard: Supporting

transitions in work through whiteboard capture and

reuse. In Proceedings of the 28th International Con-

ference on Human Factors in Computing Systems,

CHI ’10, pages 75–84. ACM.

Golovchinsky, G., Carter, S., and Biehl, J. T. (2009). Be-

yond the drawing board: Toward more effective use

of whiteboard content. CoRR, abs/0911.0039.

Ishitoya, K., Ohira, S., and Nagao, K. (2009). TimeMa-

chineBoard: A casual meeting system capable of

reusing previous discussions. Procceedings of the

International Conference on Collaboration Technolo-

gies (CollabTech2009), pages 84–89.

Moran, T. P. and van Melle, W. (2000). Tivoli: Integrat-

ing structured domain objects into a freeform white-

board environment. In CHI ’00: Extended Abstracts

on Human Factors in Computing Systems, pages 20–

21. ACM.

Oguni, T. and Nakagawa, M. (1996). Prototyping of various

applications for an interactive electronic white board.

IPSJ SIG Notes, 96(62):9–16.

Streitz, N. A., Geißler, J., Haake, J. M., and Hol, J. (1994).

Dolphin: Integrated meeting support across local and

remote desktop environments and liveboards. In

CSCW ’94: Proceedings of the 1994 ACM Con-

ference on Computer Supported Cooperative Work,

pages 345–358. ACM.

Wilcox, L. D., Schilit, B. N., and Sawhney, N. (1997).

Dynomite: A dynamically organized ink and audio

notebook. In CHI ’97: Proceedings of the SIGCHI

Conference on Human Factors in Computing Systems,

pages 186–193. ACM.

DISCOVERY OF MEETING-PARTICLE LINKS AND THEIR APPLICATION TO MEETING RECOLLECTION

SUPPORT

459