Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
Development of Micro-Blog Based DiscussionEnvironment for Participants in University Lecture
Shimpei Matsumoto,Member, IAENG,and Tomoko Kashima,Member, IAENG
Abstract—The aim of this study is to evaluate the effectivenessof micro-blog based discussion support system in dialoguesituation. Especially as a typical example of dialogue situation,this paper offers interactivity for traditional-style universitylectures including lecturer and students. To support the discus-sions, Twitter clonal Web application is developed with LAMPand Ajax, and it works as information transmission platformenhancing interactions and collaborative activity on the virtualspace. The proposal system is working during lecture, wherethe Ajax based user interface is dynamically updated in realtime. All students can post their comments and questions bythe form of short message through the web browser on eachown cell/smart phone or PC under unique handle name. Theproposal system is only available within local area network, andonly messages about the contents in the lecture are accepted byoperational guideline. The proposal system continues to workduring lecture, and it keeps the participants updated about theinteraction. The situation of discussion also keeps displaying inprojector screen or lecturer’s own PC, so the lecturer can get toknow the looks of students such as the depth of understandingand the difficulty level of the contents, and besides, students caneasily cooperate with each other. The features of the proposalsystem are designed by reference to the advantages of someprevious works on discussion mining and collaborative learning.
Index Terms—Supporting Discussion Activity, Micro-BlogBased Discussion Support, Twitter, Web, Discussion Mining,Collaborative Learning
I. I NTRODUCTION
T HE authors have previously developed an interactionsupport system between participants in lecture for pro-
moting knowledge distribution and improving reusability ofknowledge [1]. The authors have been now interested in aconspicuous characteristic of social Web typified by Twitter,interactivity and real-time property, and proposed a discus-sion support environment which positioned a website ofmicro-blog based message board as information transmissionplatform. The proposal system has been designed to beintended for lectures in university. All audiences, who arestudents in the case of university lecrtures, can post theirquestions and comments during lecture at any time when theyneed. All participants can always see the present condition ofdiscussion and also its process on the Web through browseron their own PCs or smart phones since the proposal systemautomatically continues to update the interactions in realtime. Most of all, the proposal system can entertain manystudents, so we consider that it has a potential for stimulating
Shimpei Matsumoto is with the Department of Information Systemsand Management, Faculty of Applied Information Science, HiroshimaInstitute of Technology, and also with the Graduate School of Engineering.Address: 2-1-1 Miyake, Saeki-ku, Hiroshima 731-5193, Japan. e-mail:[email protected]. URL: http://www.it-hiroshima.ac.jp/
Tomoko Kashima is with the Department of Information and SystemsEngineering, Faculty of Engineering, Kinki University. Address: 1 TakayaUmenobe, Higashi-Hiroshima City, Hiroshima, 739-2116, Japan. E-mail:[email protected]
discussions. Similarly the students’ messages can help thelecturer to teach their understandings and to schedule theprogress of his lecture since the lecturer can check obtainedquestions and comments quickly. The students can alsooffer each other advice and assistance to understand thestory of lecture, so the lecture with the proposal systemwill be more fruitful than the usual typical lecture withoutsome information and communications technology (ICT)tool. Additionally the authors have thought to be valuablefor utilizing the history data of messages stored in database.As for the technology front, the proposal system has beenimplemented by micro blog based web application withLAMP. To perform real-time property, the user interface wasdeveloped by Ajax, automatically updatable client side Webinterface technology, then it can work simultaneously withreal discussion. While taking a lecture, virtual interactionsare concurrently generated on projector screen, each one’scomputer screen, or each one’s smart phone. The authorsexpect that the virtual interactions might activate the usuallectures.
This paper examines the effectiveness of the proposalsystem for supporting lecture and its progress managementby actually introducing the real discussion situation on a trial.Specifically, the effectiveness was evaluated by a researchmeeting in laboratory. The final goal of this study is to applyfor university lecture with a lot of students. Until now, wecompleted the basis environment and operational guidelineof the proposal system. This paper reports the experimentalresults obtained by a meeting in an author’s laboratory forjunior year students. From the experimental results, we feltthat the interactivity in discussion were activating more withthe support of the proposal system.
II. RELATED WORKS AND BACKGROUND
A. Related Works
There have been many researches for supporting inter-actions in dialogue situation, and among them, developingsystems to provide knowledge sharing and to generate knowl-edge distribution have attracted much attention. The purposeand scale of dialogue situation are diverse. For example,seminars or meetings in laboratory are the case of the smallscale, on the other hand, conferences or lectures are the caseof the large one. Today for the various dialogue situations,computer based communication support systems have beendeveloped which can smoothly feedback listeners’ thinkings,opinions, comments, questions, and understandings to thelecturer or speaker, and the subject is a hot topics as researchin the field of ICT [2]. With sophistication of ICT anddissemination of cheaply-available personal digital deviceswith high-performance and multifunctions, the environmentto implement the usual ideas for supporting knowledge
Lecturer
Students with network accessible devices
Projector screen showing questions and comments
Fig. 1. Usage Example of the Proposal System
distribution is now being enhanced. As previous workspromoting knowledge sharing and distribution in dialoguesituation, an interactive lecture by sharing screen and theuse of digital pen [3], an automatic discussion developmentby voice recognition technology [4], and an acquisitiontechnique of nonverbal interaction by motion capture andcamera devices [5] are held up as the examples. Furthermoremany researches of ICT in education have dealt with lecturesand classes as dialogue situation, and they have reportedsome support methods of the interaction. As a typical ex-ample of these, we can show a report developing a lecturesupport system which can obtain questionnaires and student’sanswers from portable digital devices such as cell phones,and display these results on the sharing screen according tolecturer’s question. In recent years, these practice cases andefforts can be usually seen in journals and conferences ineducation.
B. Background
As usual, typical academic meetings such as presentationsin research conferences and university lectures have followedthe presenter-centered progress. The format based on thetypical lecture has been enough only for the organizationthat there is a presenter who has advanced knowledge ona specific field, and the presenter provides an explanationfor audiences where most of their knowledge is relatively-even. However, when audiences’ values or understandings onlearning areas have extremely diversified, the typical lectureform is insufficient. The case is also similar that the progressof lecture fully depends on discussions between participantsincluding the speaker. These two cases are prominent exam-ple requiring positive statement from audiences.
Let us think the case that there is large difference in thebasic knowledge or background of learning process betweenaudiences in lecture. This situation is already common incommunities such as recent universities because this typeof community is usually large scale, and it accepts peo-ple with various types of values. Under this situation, theprogress management of lecture and determination of lecturecontents are difficult for the lecturer without participants’statements. If the lecturer can positively obtain the responsesof participants each time during the lecture, the lecturer canimmediately reflect the obtained comments into the just lec-ture and future lectures. As the result, entire understandingsand satisfactions may be improved. Furthermore, if thereis an environment for voluntarily producing complementarycollaborative activities from participants, each participantcan help and complement each other. As the result, the
lecturer will not have to respond questions from participantsindividually, and then smooth progress of the lecture maybe possible. It is important factor that the relationship ofall participants. The answer from a participant is moreefficient than from the lecturer because the difference ofunderstandings is relatively small between participants, andpsychological hurdle for asking questions is low.
III. PROPOSAL
The system developed by this paper provides virtualcommunication opportunity similar to Twitter. The differencefrom Twitter is the usage which the proposal system isused in real dialogue situation and it is applied to onlyclosed local network environment, but Twitter is used in opennetwork environment on the Internet. Interactive informationdistribution between participants in discussions would besupported by the advantage of micro blog based socialWeb, immediacy. At present, the proposal system coversuniversity lectures as an application example of dialoguesituation. The concept of the proposal system is shown in Fig.1. The proposal system is a Web application implementedby LAMP: Linux, Apache, MySQL and PHP. Ajax basedUI of the system is displayed at exclusive screen duringlecture, so all participants in lecture can share the conditionof virtual discussion updated in real time. Each participantnecessarily brings some kind of network accessible digitaldevice into the lecture, and the network usage is availablein the lecture room everywhere without regard for wired orwireless. This is an important precondition of the proposalsystem. Lecture rooms in the authors’ institute have twoprojector screens, and different contents can be displayedeach other. Therefore for example, the proposal system isthought to be used by that the one screen displays UI of theproposal system, and the other displays lecture materials.Even under the environment with only one projector screen,the proposal system is also available by using extensiondisplay setting, which can display different contents betweenPC and projector screen like multi monitor. In this condition,the lecturer’s own PC monitor shows the proposal system,and the projector screen shows lecture materials.
As previous arrangement to operate the proposal system,each participant accesses to the Web server through Webbrowser and login to the system under unique ID. Allparticipants make the system operational with active windowin accordance with the lecturer’s direction, so the condition ofdiscussion continually-updated is always available from eachparticipant’s device including the lecturer. The participantsmay freely post their message by inputting the message tothe posting form during lecture, and the result is instantlyreflected to the UI owing to Ajax with jQuery. With theproposal system, interactive information distribution will beactivated, and comments and questions from the audienceswill be sent to the lecturer instantly and casually. Not only theopportunity to send messages from audiences to lecturer, thisstudy expects that collaborative activity between audienceswill be easily generated during lecture. The lecturer proceedsthe discussion with checking UI of the proposal systemin the interval of lecture. If an interesting comment or animportant question is posted, the lecturer can respond or canadd the explanation about this message quickly. Also themessages may be useful for adjusting the lecture speed and
37.1%
40.0%
20.0%
2.9%
Comments
Questions
Notes
Collaborations
Fig. 2. Rates of Posts in All Presentations
making lecture materials because the contents of messagesare related with the degrees of understanding. The more thesystem is used, the more the log data is automatically stored,and obtained log data is thought to be extremely valuablebecause it will be available for data analysis. Lecturer canpursue his/her talk with real time feedback from audiences’comments, further interactivity is expected more than theusual format of university lectures.
IV. EXPERIMENT
A. Experimental Conditions
This section shows experimental conditions. The proposalsystem is designed for applying real lecture, but this paperused it for a small dialogue situation and evaluates its effec-tiveness because the main work of this paper is preparing forintroduction of the proposal system into lectures. Specificallythe proposal system was applied for an annual researchmeeting in laboratory. This paper expects the members inlaboratory to give much data because the relation of membersin laboratory is closer than the relation of audiences inlecture, so psychological hurdle for posts may be relativelylow. One university teaching staff (academic adviser) and tenjunior year students joined in the experiment. The meetingtheme was senior year students’ research subjects reportedfrom the interim presentation held a few days before. Inthe experiment, each research subject was discussed withthe abstract papers. Two experiments were conducted once aweek, and each experimental time was 60 minutes. In the firstexperiment, the academic adviser concentrated on speakingall research subjects, and did not post. The participants wereinstructed to freely post their honest opinion during thepresentation. Some examples of messages were previouslyshown, such as an honest feedback about the subject, aproposal which can improve the idea more, a questionabout difficulty point, and a negative opinion. There wasno restriction on the number of posts and its content, andall messages followed individual initiative. In the secondexperiment, each student gave a presentation on a researchsubject he/her interested in, and at that time, the academicadviser joined posting together with other students. 9 ofthe audiences were with PCs, and a audience was with asmart phone. The first experiment gave each presentation 15minutes where there were 4 presentations, and the secondexperiment gave each presentation 10 minutes where therewere 6 presentations.
B. Results and Discussion
Each experiment obtained the average of 257 posts. Eachpost was classified into 4 groups according to its content by
(a) Type A
(b) Type B
data processing engineer
examination
LMS
learner efficient
advance
development
data
analysis
databasestructure
investigation
database
have
function forum
lecture
practical
verification
research
effect
exercises
use
understanding
do
knowledge processing
information system modeling
effect ofstudy
question
course
solve
tabase
qu
purpose
registration
datata
MoodleM
practicalpractical
aminat
ner
redadadadadadadadadadatatatatatatata
ququ
analysysis
utilize
lecture
investigation
material
author 2 saddle
performance of competition
lower-limb
muscle
author 1
sasaddlesadd
approval
give
exercise
improperutilize
bicycle
evaluate
use
confirmation
excel
utilize
high of saddle
research
presentation
information
effective
myoelectric purpose
analysis
difference
continuation
change of wavelet
method of pattern entropy
electric potential
surface
qualitative change
application
se
impr
nfirmaric
ututilililililililizizizizizizizize
pe
saddle
Fig. 3. Network Structures of 2 Types of Presentations
the authors’ vision.• Comments: Thinkings, statements, proposals and ob-
jections.• Questions: Questions about exploration.• Notes: Posts reciting important points and provisions of
related information.• Collaborations: Questions for an audience’s post, and
answers to an audience’s question.The rates of 4 groups are shown in Fig. 2. Collaborative
activities were fewer than originally expected, but the rateof collaborative activities is over 50 % if the provisionsof related information and the proposals are classified intothe collaborations. Therefore we can consider that the in-teractions between participants was enough supported. Theresult provides a future work, development of an automaticclassification method based on relationships between words.
The tendencies of the posts depending on the contents ofpresentation were examined by a network analysis method.In this analysis, words in each abstract paper of presentationwere used. Relational strength between words was calculatedbased on the probability and the frequency of appearance ofwords in an abstract paper, and then a word-network diagramwas generated where two words with strong relationship wereconnected. We can visually understand the characteristic ofpresentation from the word-network diagram. Two typicalexamples are shown in Fig. 3. The size of nodes and thewidth of links between nodes correspond to the appearancefrequency of words. The word-network was generated byextracting pairs of nouns, verbs, and Japanese specific irreg-ular conjunctive nouns (Sa-gyo irregular conjugation). In theword-network, each pair of words forms some clusters. Forexample in Fig. 3 (a), there is a large cluster. In contract,Fig. 3 (b) has some clusters. The cluster itself is assumed tobe a topic because the cluster is a group of words spokentogether. So Fig. 3 (a) is a presentation with one large topic.This paper discussed word-network diagrams in relation to
0%
20%
40%
60%
80%
100%
All Type A Type B
Collaborations
Notes
Questions
Comments
Fig. 4. Rates of Posts depending on the Type of Presentation
the contents of posts. In the presentation with one core topic(large cluster) like Fig. 3 (a), there was a tendency that a lotof proposals about the presentation or each one’s opinionswere received. On the other hand, in the presentation withsome small clusters like Fig. 3 (b), there was a tendency thata lot of notes and questions were received. To ensure theresults and discussion in Fig. 3, the rates of posts dependingon the word-network structure was made as shown in Fig. 4.From Fig. 4, the contents of posts were noticeably differentdepending on the word-network structure.
Overall tendency is as follows. The number of posts wasrelatively few in the presentation whose objective was easy-to-understand or detail was difficult. On the other hand,the number of posts was larger in the presentation whoseexplanation of preconditions was insufficient. In the presen-tation easily understandable, there were many posts whichhad nothing to do with the progress of presentation. Whenspeaker shows many concrete examples in the presentation,many posts were received which deeply related with theprogress of presentation. This tendency was similar in thepresentation which was conveniently added some supplemen-tal explanations. As mentioned above, the number of postsand their contents were notably changed depending on theformat of presentation.
The number of characters of all messages in a dialogueopportunity was examined. Fig. 5 shows the frequency dis-tribution depending on the number of Japanese characters.We can see that most messages were with short sentencefrom Fig. 5. The average of characters is approximately20, and there were few messages with over 50 characters.After the experiment, the proposal system was subjectivelyevaluated by the interviews from speakers and audiences.According to speakers, it was enough to understand theaudiences’ points even messages with short sentence, so thespeakers could timely add the exploration in response to themessages. The proposal system was evaluated as effectivediscussion support by the speakers. Additionally, most ofaudiences said that short sentence based post is convenientbecause the input work of short messages from keyboard isnot a heavy burden even while listening presentation. Andalso there was an opinion that the meeting with the proposalsystem was more exciting than the usual because the messageboard was frequently updated owing to the short sentence,then the virtual discussion could be developed along with theprogress of real presentation. Therefore this paper concludedthat the proposal system is effective as discussion support forall participants in dialogue situation.
0
5
10
15
20
25
30
1 16 31 46
Degrees
Number of
Characters
Fig. 5. Frequency Distribution depending on the Number of Characters
V. CONCLUSION
This paper established a micro blog based informationtransmission platform for university lectures to provide in-teractivity and collaboration opportunity between participantsincluding speaker and audiences. This paper conducted someexperiments to confirm whether the proposal system cangenerate opportunity for questions and comments to thespeaker, and can promote audience-to-audience communi-cations. From the experiments, the proposal system wasefficient for supporting the progress of discussion for speaker.Most of obtained messages were 50 characters or less,but audiences give us many feedbacks that short sentencebased post is the most suitable because the audiences haveto post while listening presentation. Each message was anadequate amount of characters for speakers to check whilepresentation, and was enough to understand the point ofdiscussion even short sentence. Speaker could feed someimportant points back into the presentation immediately. Asthis paper had applied the proposal system for the smalllecture, the proposal system is thought to be effective aslecture support method because students’ comments andquestions were directly obtained during lecture.
In the near future, the proposal system will be introducedon a trial basis to a large lecture. In the authors’ faculty,all students have a laptop PC, and they bring each own PCto specialized lectures in accordance with the direction oflecturer. Therefore there is probably no barrier for studentsto use the proposal system. Currently we have no firmoperational guideline, so it will be developed through thetrials to obtain sufficient and suitable comments.
REFERENCES
[1] S. Matsumoto, Y. Kawamoto, T. Kashima and K. Funamoto, “EvaluatingMicro-Blog based Message Board for Supporting Discussion Activityin Dialogue Situation,”The Papers of Technical Meeting on InformationSystems, IEE Japan, IS-11-075-092, pp.51-56,2011 (In Japanese).
[2] K. Nagao, k. Kaji, D. Yamamoto and H. Tomobe, “Discussion Mining:Annotation-Based Knowledge Discovery from Real World Activities,”Proc. of the Fifth Pacific-Rim Conference on Multimedia, Part 1, pp.522-531, 2004.
[3] T. Sugihara, M. Miura and S. Kunito, “Visualizing the Process ofLesson with Digital Pen System and Increasing Interaction,”Proc. ofInteraction 2010, pp.135-142, 2010 (In Japanese).
[4] U. Cho, M. Matsumura and M. Taniuchida, “Auto-Generation ofDiscussion-Map from Speech-Recognized Minute,”Proc. of the 20thAnnual Conference of Japanese Society for Artificial Intelligence, 3C3-5, 2006 (In Japanese).
[5] H. Kachigi and Y. Kaku, “A Meeting Capture System based on Partic-ipants’ Nonverbal Interaction and Collaborative Annotation,”Proc. ofInteraction 2011, 1CR3-6, 2011 (In Japanese).