

 Kobe University Repository : Kernel  

タイトル
Tit le

Analyzing Learning Pat t erns Based on Log Dat a from Digit al
Text books

著者
Aut hor(s)

Mouri, Kousuke / Ren, Zhuo / Uosaki, Noriko / Yin, Chengjiu

掲載誌・巻号・ページ
Cit at ion

Int ernat ional Journal of Dist ance Educat ion Technologies
(IJDET),17(1):1-14

刊行日
Issue dat e

2019-03

資源タイプ
Resource Type

Journal Art icle / 学術雑誌論文

版区分
Resource Version

publisher

権利
Right s

DOI 10.4018/IJDET.2019010101

JaLCDOI

URL ht t p://www.lib.kobe-u.ac.jp/handle_kernel/90006335

PDF issue: 2021-04-06


DOI: 10.4018/IJDET.2019010101

International Journal of Distance Education Technologies
Volume 17 • Issue 1 • January-March 2019

﻿
Copyright﻿©﻿2019,﻿IGI﻿Global.﻿Copying﻿or﻿distributing﻿in﻿print﻿or﻿electronic﻿forms﻿without﻿written﻿permission﻿of﻿IGI﻿Global﻿is﻿prohibited.

﻿

1

Analyzing Learning Patterns Based 
on Log Data from Digital Textbooks
Kousuke Mouri, Tokyo University of Agriculture and Technology, Fuchu, Japan

Zhuo Ren, Jinan University, Guangzhou, China

Noriko Uosaki, Osaka University, Suita, Japan

Chengjiu Yin, Kobe University, Kobe, Japan

ABSTRACT

The﻿analysis﻿of﻿learning﻿behaviors﻿from﻿the﻿log﻿data﻿of﻿digital﻿textbooks﻿is﻿beneficial﻿for﻿improving﻿
education﻿systems.﻿The﻿focus﻿of﻿discussion﻿in﻿any﻿analysis﻿of﻿learning﻿behaviors﻿is﻿often﻿on﻿discovering﻿
the﻿relationships﻿between﻿learning﻿behavior﻿and﻿learning﻿performance.﻿However,﻿little﻿attention﻿has﻿
been﻿ paid﻿ to﻿ investigating﻿ and﻿ analyzing﻿ learning﻿ patterns﻿ or﻿ rules﻿ among﻿ learning﻿ style﻿ of﻿ index﻿
(LSI),﻿ cognitive﻿ style﻿ of﻿ index﻿ (CSI),﻿ and﻿ the﻿ logs﻿ of﻿ digital﻿ textbooks.﻿ In﻿ this﻿ study,﻿ the﻿ authors﻿
proposed﻿a﻿method﻿to﻿analyze﻿learning﻿patterns﻿or﻿rules﻿of﻿reading﻿digital﻿textbooks.﻿The﻿analysis﻿
method﻿used﻿association﻿analysis﻿with﻿the﻿Apriori﻿algorithm.﻿The﻿analysis﻿was﻿conducted﻿using﻿logs﻿
of﻿digital﻿textbooks﻿and﻿questionnaires﻿to﻿investigate﻿students’﻿learning﻿and﻿cognitive﻿styles.﻿From﻿
the﻿detected﻿meaningful﻿association﻿rules,﻿this﻿study﻿found﻿three﻿student﻿types:﻿poorly﻿motivated,﻿
efficient,﻿and﻿diligent.﻿The﻿authors﻿believe﻿that﻿consideration﻿of﻿these﻿student﻿types﻿can﻿contribute﻿
to﻿the﻿improvement﻿of﻿learning﻿and﻿teaching

KEywoRDS
Association Rule, Cognitive Style, Digital Textbooks Reading Log, Learning Analytics, Learning Style

INTRoDUCTIoN

With﻿the﻿development﻿of﻿e-publishing﻿technologies﻿and﻿standards,﻿it﻿is﻿easy﻿to﻿obtain﻿digital﻿books,﻿
such﻿as﻿“living﻿books,”﻿“talking﻿books,”﻿and﻿“CD-ROM﻿books,”﻿from﻿the﻿Internet﻿(Yin﻿et﻿al.,﻿2015).﻿
Digital﻿books﻿have﻿become﻿a﻿potentially﻿effective﻿pedagogic﻿tool﻿(Hezroni,﻿2004;﻿Reinking,﻿1997;﻿
Snyder,﻿2002),﻿indicated﻿by﻿the﻿fact﻿that﻿digital﻿book﻿reading﻿has﻿increased﻿significantly﻿in﻿the﻿United﻿
States﻿(Lee﻿et﻿al.,﻿2012).﻿Consequently,﻿traditional﻿textbooks﻿are﻿being﻿increasingly﻿replaced﻿by﻿digital﻿
textbooks﻿(Ren,﻿Uosaki,﻿Kumamoto,﻿Liu,﻿&﻿Yin,﻿2017).

Many﻿ researchers﻿ have﻿ been﻿ paying﻿ attention﻿ to﻿ the﻿ development﻿ of﻿ digital﻿ books﻿ to﻿ support﻿
teaching,﻿learning,﻿and﻿scholarship.﻿By﻿using﻿digital﻿books,﻿large﻿bodies﻿of﻿log﻿data﻿can﻿be﻿accumulated,﻿
the﻿analysis﻿of﻿which﻿can﻿be﻿used﻿to﻿perform﻿learning﻿analytics.

Learning﻿analytics﻿can﻿be﻿used﻿in﻿making﻿suggestions﻿to﻿policy﻿makers,﻿instructors,﻿and﻿learners﻿
(Baker﻿&﻿Inventado,﻿2014;﻿Hwang,﻿Hsu,﻿Lai,﻿&﻿Hsueh,﻿2017).﻿Therefore,﻿learning﻿analytics﻿have﻿
become﻿an﻿important﻿issue﻿in﻿education﻿(Hwang,﻿Chu,﻿&﻿Yin,﻿2017)﻿that﻿have﻿entailed﻿important﻿
changes﻿in﻿educational﻿research.


International Journal of Distance Education Technologies
Volume 17 • Issue 1 • January-March 2019

2

The﻿ objective﻿ of﻿ learning﻿ analytics﻿ is﻿ to﻿ provide﻿ helpful﻿ information﻿ to﻿ optimize﻿ or﻿ improve﻿
learning﻿designs,﻿learning﻿outcomes,﻿and﻿learning﻿environments﻿based﻿on﻿the﻿analysis﻿results﻿(Greller﻿
&﻿Drachsler,﻿2012;﻿Hwang,﻿Chu,﻿&﻿Yin,﻿2017).

In﻿the﻿analysis﻿of﻿learning﻿behaviors﻿in﻿this﻿study,﻿we﻿used﻿a﻿digital﻿textbook﻿system﻿to﻿collect﻿
students’﻿learning﻿logs.﻿Learning﻿log﻿is﻿defined﻿as﻿a﻿digital﻿record﻿of﻿what﻿learners﻿have﻿learned﻿in﻿a﻿
formal﻿and﻿an﻿informal﻿setting﻿(Ogata,﻿Hou,﻿Uosaki,﻿Mouri,﻿&﻿Liu.,﻿2014;﻿Mouri,﻿Ogata,﻿&﻿Uosaki,﻿
2015).﻿ The﻿ system﻿ was﻿ used﻿ in﻿ a﻿ commercial﻿ law﻿ course﻿ for﻿ undergraduate﻿ students,﻿ which﻿ was﻿
conducted﻿in﻿entirety﻿in﻿English:﻿The﻿students﻿were﻿assigned﻿readings﻿in﻿English,﻿and﻿the﻿teacher﻿spoke﻿
in﻿English.﻿We﻿also﻿used﻿questionnaires﻿to﻿collect﻿data﻿on﻿students’﻿learning﻿styles﻿and﻿cognitive﻿styles.

Using﻿these﻿data,﻿we﻿applied﻿the﻿association﻿analysis﻿method﻿with﻿the﻿Apriori﻿algorithm﻿to﻿analyze﻿
students’﻿learning﻿patterns﻿or﻿rules.﻿One﻿of﻿the﻿advantages﻿of﻿analyzing﻿the﻿learning﻿patterns﻿or﻿rules﻿
is﻿the﻿preemptive﻿prediction﻿of﻿students’﻿final﻿grade﻿and﻿progresses﻿in﻿the﻿future.﻿As﻿a﻿result,﻿teachers﻿
can﻿improve﻿their﻿teaching﻿strategies﻿and﻿support﻿students’﻿learning﻿behaviors.﻿From﻿the﻿analysis,﻿this﻿
study﻿found﻿three﻿meaningful﻿student﻿types﻿by﻿considering﻿the﻿detected﻿association﻿rules.

LITERATURE REVIEw

Previous Studies of Data Collection
Data﻿collection﻿is﻿the﻿first﻿step﻿in﻿learning﻿analysis﻿(Yin,﻿Hirokawa,﻿et﻿al.,﻿2013;﻿Yin,﻿Sung,﻿et﻿al.,﻿
2013).﻿ Yin﻿ et﻿ al.﻿ (2016)﻿ performed﻿ a﻿ review﻿ of﻿ previous﻿ research﻿ to﻿ survey﻿ the﻿ methods﻿ of﻿ data﻿
collection.﻿Based﻿on﻿the﻿data﻿collection﻿source,﻿previous﻿studies﻿of﻿data﻿collection﻿can﻿be﻿classified﻿
into﻿three﻿types:﻿Questionnaire-Based﻿Data﻿Collection﻿(QDC),﻿Manual﻿Data﻿Collection﻿(MDC),﻿and﻿
Automatic﻿Data﻿Collection﻿(ADC)﻿(Yin﻿et﻿al.,﻿2014;﻿Yin﻿et﻿al.,﻿2017).

•﻿ QDC.﻿In﻿this﻿method,﻿data﻿are﻿collected﻿by﻿using﻿a﻿predesigned﻿questionnaire.﻿The﻿questionnaire﻿
is﻿a﻿tool﻿to﻿elicit﻿specific﻿responses﻿from﻿the﻿participants﻿of﻿the﻿survey,﻿and﻿it﻿is﻿a﻿data-gathering﻿
method﻿used﻿to﻿collect﻿and﻿analyze﻿the﻿feedback﻿of﻿a﻿group﻿of﻿people﻿from﻿a﻿target﻿population.

•﻿ MDC.﻿In﻿this﻿method,﻿a﻿manual﻿data﻿collection﻿system﻿is﻿open﻿to﻿users﻿of﻿the﻿system﻿to﻿consciously﻿
provide﻿data﻿about﻿their﻿learning﻿behaviors.﻿Users﻿may﻿save﻿data﻿that﻿they﻿consider﻿useful﻿through﻿
the﻿system﻿by﻿themselves.﻿For﻿example,﻿if﻿a﻿student﻿encounters﻿interesting﻿images﻿that﻿he﻿or﻿she﻿
wants﻿to﻿share﻿with﻿his﻿or﻿her﻿friends,﻿then﻿he﻿or﻿she﻿can﻿capture﻿images﻿from﻿an﻿authentic﻿and﻿
shareable﻿environment.

•﻿ ADC.﻿In﻿this﻿method,﻿students’﻿log﻿data﻿for﻿learning﻿behaviors﻿are﻿automatically﻿recorded﻿while﻿
reading﻿ e-documents.﻿ For﻿ example,﻿ Yin﻿ et﻿ al.﻿ (2015)﻿ analyzed﻿ learning﻿ behavior﻿ to﻿ identify﻿
students’﻿learning﻿style﻿using﻿data﻿from﻿the﻿automatically﻿recorded﻿reading﻿logs﻿of﻿the﻿students’﻿
digital﻿ textbooks.﻿ By﻿ using﻿ the﻿ same﻿ digital﻿ textbook﻿ log﻿ data,﻿ Shimada,﻿ Okubo,﻿ Yin,﻿ and﻿
Ogata﻿(2017)﻿summarized﻿lecture﻿slides﻿to﻿enhance﻿preview﻿efficiency﻿and﻿improve﻿students’﻿
understanding﻿ of﻿ the﻿ content,﻿ and﻿ Mouri﻿ and﻿ Yin﻿ (2017)﻿ found﻿ some﻿ patterns﻿ for﻿ improving﻿
learning﻿materials.

In﻿the﻿present﻿work,﻿we﻿used﻿QDC﻿and﻿ADC﻿methods﻿to﻿collect﻿data.﻿Three﻿kinds﻿of﻿data﻿were﻿
used﻿to﻿analyze﻿students’﻿learning﻿patterns.﻿The﻿first﻿was﻿log﻿data﻿from﻿digital﻿textbooks﻿(ADC).﻿The﻿
second﻿was﻿students’﻿learning﻿style﻿data﻿collected﻿through﻿the﻿Index﻿of﻿Learning﻿Styles﻿questionnaire.﻿
Felder﻿and﻿Soloman﻿(2001)﻿developed﻿this﻿questionnaire﻿to﻿assess﻿preferences﻿on﻿four﻿dimensions﻿
(active/reflective,﻿sensing/intuitive,﻿visual/verbal,﻿and﻿sequential/global)﻿of﻿a﻿learning﻿style﻿model.﻿
The﻿third﻿was﻿students’﻿cognitive﻿style﻿data,﻿collected﻿with﻿the﻿Cognitive﻿Style﻿Index﻿questionnaire.﻿
Allinson﻿and﻿Hayes﻿(1996)﻿developed﻿this﻿questionnaire,﻿which﻿is﻿widely﻿used﻿to﻿measure﻿cognitive﻿
styles﻿in﻿the﻿field﻿of﻿education.


International Journal of Distance Education Technologies
Volume 17 • Issue 1 • January-March 2019

3

Digital Textbooks
In﻿the﻿past﻿decade,﻿various﻿studies﻿have﻿been﻿conducted﻿to﻿investigate﻿the﻿effectiveness﻿of﻿learning﻿
with﻿digital﻿textbooks.﻿For﻿example,﻿Shepperd,﻿Grace,﻿and﻿Koch﻿(2008)﻿compared﻿the﻿efficacy﻿of﻿
digital﻿textbooks﻿and﻿traditional﻿textbooks﻿and﻿indicated﻿that﻿students﻿rated﻿the﻿usability﻿of﻿digital﻿
textbooks﻿positively.﻿Rockinson-Szapkiw,﻿Courduff,﻿Carter,﻿and﻿Bennett﻿(2013)﻿compared﻿the﻿learning﻿
effectiveness﻿of﻿digital﻿textbooks﻿and﻿traditional﻿textbooks﻿and﻿found﻿that﻿digital﻿textbooks﻿are﻿as﻿
effective﻿for﻿learning﻿as﻿traditional﻿textbooks.

In﻿ contrast﻿ to﻿ traditional﻿ textbooks,﻿ digital﻿ textbooks﻿ can﻿ offer﻿ digital﻿ listening,﻿ reading,﻿ and﻿
vocabulary﻿practice.﻿Therefore,﻿it﻿is﻿necessary﻿to﻿consider﻿the﻿design﻿of﻿digital﻿textbooks﻿for﻿offering﻿
effective﻿learning.﻿Gu,﻿Wu,﻿and﻿Xu﻿(2015)﻿reported﻿the﻿importance﻿regarding﻿the﻿design﻿of﻿digital﻿
textbooks﻿and﻿suggested﻿that﻿well-designed﻿digital﻿textbooks﻿positively﻿enhance﻿learning.

Therefore,﻿many﻿researchers﻿concur﻿that﻿digital﻿textbooks﻿have﻿become﻿a﻿potentially﻿effective﻿
pedagogic﻿ tool﻿ to﻿ support﻿ teaching,﻿ learning,﻿ and﻿ scholarship﻿ (Hezroni,﻿ 2004;﻿ Reinking,﻿ 1997;﻿
Snyder,﻿2002).

A﻿large﻿body﻿of﻿log﻿data﻿can﻿be﻿accumulated﻿using﻿digital﻿textbook﻿systems﻿for﻿the﻿purpose﻿of﻿
monitoring﻿students’﻿activities.﻿However,﻿there﻿are﻿few﻿studies﻿analyzing﻿the﻿relationships﻿between﻿
learners’﻿ learning﻿ style,﻿ cognitive﻿ style,﻿ and﻿ the﻿ log﻿ data﻿ from﻿ digital﻿ textbooks.﻿ We﻿ believe﻿ that﻿
analyzing﻿such﻿relationships﻿can﻿help﻿in﻿providing﻿different﻿forms﻿of﻿effective﻿learning﻿support﻿in﻿
accordance﻿with﻿their﻿learning﻿and﻿cognitive﻿styles.

Association Analysis
This﻿study﻿employed﻿association﻿analysis﻿using﻿the﻿Apriori﻿algorithm.﻿This﻿method﻿is﻿designed﻿to﻿
extract﻿association﻿rules﻿from﻿a﻿database﻿containing﻿transactions,﻿such﻿as﻿collections﻿of﻿items﻿bought﻿
by﻿customers﻿or﻿details﻿of﻿website﻿frequentation.﻿In﻿educational﻿technology﻿fields,﻿researchers﻿focus﻿
on﻿this﻿method﻿of﻿analysis﻿to﻿mine﻿regularities﻿among﻿some﻿parameters﻿of﻿educational﻿big﻿data.

For﻿example,﻿Behrouz,﻿Gerd,﻿and﻿William﻿(2004)﻿found﻿association﻿rules﻿by﻿grouping﻿students﻿
who﻿were﻿enrolled﻿in﻿an﻿online﻿education﻿system﻿based﻿on﻿parameters﻿such﻿as﻿GPA﻿(Grade﻿Point﻿
Average),﻿age,﻿and﻿gender.﻿In﻿their﻿study,﻿if﻿there﻿were﻿students﻿with﻿GPA﻿scores﻿between﻿3.0﻿and﻿
3.5,﻿the﻿system﻿can﻿give﻿them﻿the﻿probability﻿of﻿whether﻿they﻿can﻿pass﻿a﻿course﻿that﻿they﻿will﻿attend﻿
based﻿on﻿the﻿detected﻿association﻿rules.

Mouri﻿ and﻿ colleagues﻿ (Mouri,﻿ Ogata,﻿ &﻿ Uosaki,﻿ 2016;﻿ Mouri,﻿ Okubo,﻿ Shimada,﻿ &﻿ Ogata,﻿
2016)﻿used﻿association﻿analysis﻿to﻿mine﻿useful﻿rules﻿or﻿patterns﻿from﻿learning﻿logs﻿accumulated﻿in﻿
a﻿ ubiquitous﻿ learning﻿ system.﻿ By﻿ providing﻿ them﻿ advice﻿ based﻿ on﻿ the﻿ detected﻿ association﻿ rules,﻿
students’﻿learning﻿activities﻿in﻿informal﻿settings﻿can﻿be﻿improved.﻿However,﻿little﻿attention﻿has﻿been﻿
paid﻿to﻿analyzing﻿data﻿such﻿as﻿Learning﻿Style﻿Index﻿(LSI)﻿and﻿Cognitive﻿Style﻿Index﻿(CSI)﻿to﻿find﻿
the﻿association﻿rules﻿between﻿learning﻿styles﻿and﻿digital﻿textbook﻿logs.﻿By﻿analyzing﻿the﻿relations﻿
between﻿learning﻿styles﻿and﻿digital﻿textbook﻿logs,﻿there﻿is﻿a﻿possibility﻿that﻿we﻿can﻿identify﻿important﻿
association﻿rules﻿to﻿predict﻿future﻿learners’﻿grade﻿based﻿on﻿their﻿learning﻿styles﻿and﻿digital﻿textbook﻿
logs.﻿ Therefore,﻿ this﻿ study﻿ focuses﻿ on﻿ analyzing﻿ logs﻿ collected﻿ in﻿ a﻿ digital﻿ textbook﻿ system﻿ in﻿
combination﻿with﻿results﻿from﻿the﻿learning-style﻿or﻿cognitive-style﻿questionnaires.

SySTEM

The Architecture and Interface of the Digital Textbook System
The﻿ server﻿ side﻿ runs﻿ on﻿ CentOS,﻿ and﻿ it﻿ is﻿ programmed﻿ using﻿ Java﻿ and﻿ Mysql.﻿ The﻿ client﻿ side﻿ is﻿
working﻿ on﻿ a﻿ web﻿ browser﻿ using﻿ HTML5﻿ and﻿ javascript.﻿ The﻿ users﻿ can﻿ register﻿ and﻿ read﻿ digital﻿
textbooks﻿anytime﻿and﻿anywhere.

A﻿ web-based﻿ digital﻿ textbook﻿ system﻿ using﻿ the﻿ e-pub﻿ format﻿ was﻿ developed﻿ for﻿ use﻿ in﻿ this﻿
research.﻿ This﻿ digital﻿ textbook﻿ system﻿ was﻿ developed﻿ to﻿ collect﻿ data﻿ from﻿ classes.﻿ The﻿ system﻿ is﻿


International Journal of Distance Education Technologies
Volume 17 • Issue 1 • January-March 2019

4

named﻿Digital﻿textbook﻿for﻿Improving﻿Teaching﻿and﻿Learning﻿(DITeL).﻿The﻿DITeL﻿system﻿can﻿be﻿
used﻿not﻿only﻿on﻿personal﻿computers,﻿but﻿also﻿on﻿smart﻿phones.﻿Specifically,﻿this﻿digital﻿system﻿can﻿
be﻿used﻿anywhere﻿and﻿anytime.﻿Teachers﻿and﻿students﻿can﻿use﻿the﻿DITeL﻿system﻿and﻿read﻿a﻿digital﻿
textbook﻿on﻿mobile﻿devices﻿such﻿as﻿iPads,﻿iPhones,﻿and﻿Android.﻿In﻿addition,﻿their﻿learning﻿logs﻿were﻿
collected﻿to﻿analyze﻿their﻿learning﻿behaviors﻿to﻿improve﻿the﻿DITeL﻿system.

Figure﻿1﻿and﻿Figure﻿2﻿show﻿the﻿interface﻿for﻿students﻿and﻿teachers,﻿respectively.﻿By﻿using﻿this﻿
online﻿ digital﻿ textbook﻿ reading﻿ system,﻿ we﻿ can﻿ collect﻿ data﻿ like﻿ “turning﻿ to﻿ next/previous﻿ page,”﻿
“memo,”﻿“zoom﻿in/out,”﻿and﻿“adding﻿marker.”﻿All﻿of﻿these﻿actions﻿are﻿stored﻿to﻿the﻿database.﻿These﻿
data﻿were﻿used﻿to﻿analyze﻿learning﻿behaviors.

Turning﻿to﻿next/previous﻿page.﻿Students﻿can﻿read﻿the﻿teaching﻿content﻿repeatedly;﻿they﻿can﻿go﻿
to﻿the﻿next﻿page﻿by﻿clicking﻿the﻿“Next”﻿button,﻿and﻿backtrack﻿to﻿the﻿previous﻿page﻿by﻿clicking﻿the﻿
“Prev”﻿button.

Memo.﻿When﻿a﻿user﻿wants﻿to﻿make﻿a﻿memo﻿on﻿the﻿learning﻿content,﻿he﻿or﻿she﻿clicks﻿the﻿“Memo”﻿
button,﻿which﻿shows﻿a﻿textbox.﻿After﻿the﻿memo﻿is﻿written,﻿the﻿action﻿name﻿will﻿be﻿saved﻿as﻿“Memo.”

Zoom﻿in/out.﻿The﻿zoom﻿in/out﻿function﻿can﻿help﻿students﻿read﻿the﻿contents﻿more﻿clearly.
Adding﻿marker.﻿When﻿a﻿user﻿wants﻿to﻿highlight﻿some﻿text﻿in﻿the﻿learning﻿content,﻿he﻿or﻿she﻿will﻿

click﻿the﻿“abc﻿highlight”﻿or﻿“Underline”﻿button,﻿and﻿the﻿action﻿name﻿will﻿be﻿saved﻿as﻿“Highlight”﻿
or﻿“Underline.”

The﻿teacher﻿can﻿register﻿each﻿student’s﻿name﻿and﻿number﻿into﻿the﻿system.﻿Before﻿the﻿students﻿
log﻿into﻿the﻿system,﻿the﻿digital﻿textbook﻿and﻿other﻿relevant﻿materials﻿are﻿uploaded﻿to﻿the﻿system﻿by﻿

Figure 1. Student interface of DITeL


International Journal of Distance Education Technologies
Volume 17 • Issue 1 • January-March 2019

5

the﻿teacher.﻿As﻿shown﻿in﻿Figure﻿2,﻿when﻿the﻿teacher﻿clicks﻿the﻿“Textbook﻿List”﻿button,﻿a﻿“Textbook﻿
List”﻿window﻿will﻿appear,﻿and﻿when﻿the﻿teacher﻿clicks﻿the﻿“Registration”﻿button,﻿a﻿“Textbook﻿Upd”﻿
window﻿will﻿appear﻿to﻿teacher.﻿At﻿the﻿end,﻿the﻿teacher﻿can﻿select﻿and﻿upload﻿the﻿teaching﻿materials﻿
into﻿the﻿system.

Each﻿student﻿will﻿have﻿an﻿individual﻿account﻿to﻿enter﻿into﻿the﻿system,﻿so﻿that﻿a﻿separate﻿record﻿
is﻿kept﻿for﻿the﻿learning﻿activities﻿in﻿this﻿course.

Log Data from the DITel System
The﻿data﻿were﻿collected﻿from﻿the﻿DITeL﻿system.﻿Table﻿1﻿shows﻿a﻿sample﻿of﻿reading﻿action﻿logs.﻿One﻿
data﻿log﻿contains﻿the﻿date,﻿time,﻿user﻿ID,﻿learning﻿content﻿ID,﻿page﻿number,﻿user﻿action,﻿and﻿other﻿data.

Participation
The﻿DITeL﻿system﻿was﻿used﻿in﻿a﻿commercial﻿law﻿course﻿for﻿undergraduate﻿students.﻿This﻿course﻿was﻿
conducted﻿wholly﻿in﻿English:﻿The﻿students﻿were﻿assigned﻿English﻿readings,﻿and﻿the﻿teacher﻿spoke﻿in﻿
English.﻿A﻿total﻿of﻿50,000﻿records﻿were﻿gathered﻿from﻿March﻿to﻿July﻿2017.

Figure 2. Teacher interface of DITeL


International Journal of Distance Education Technologies
Volume 17 • Issue 1 • January-March 2019

6

A﻿ total﻿ of﻿ 41﻿ undergraduate﻿ students﻿ participated﻿ in﻿ this﻿ study.﻿ The﻿ participants﻿ were﻿ asked﻿
to﻿read﻿certain﻿learning﻿content﻿(272﻿pages)﻿via﻿the﻿digital﻿textbook﻿system.﻿The﻿mean﻿age﻿of﻿the﻿
participants﻿was﻿20﻿years.

The﻿confidentiality﻿of﻿the﻿participants﻿was﻿protected﻿by﻿hiding﻿their﻿personal﻿information﻿during﻿
the﻿research﻿process;﻿moreover,﻿they﻿knew﻿that﻿their﻿participation﻿was﻿voluntary﻿and﻿that﻿they﻿could﻿
withdraw﻿from﻿the﻿study﻿at﻿any﻿time.

ANALySIS METHoD

The﻿association﻿analysis﻿was﻿conducted﻿using﻿the﻿“arules”﻿package﻿(2017)﻿of﻿the﻿R﻿language﻿and﻿
transaction﻿ data﻿ based﻿ on﻿ digital﻿ textbook﻿ logs,﻿ LSI,﻿ and﻿ CSI﻿ questionnaires.﻿ Table﻿ 2﻿ shows﻿ the﻿
ranking﻿that﻿was﻿established﻿from﻿the﻿individual﻿students’﻿reading﻿times﻿based﻿on﻿the﻿total﻿time﻿of﻿
the﻿page﻿flipping﻿logs.﻿The﻿mean﻿and﻿median﻿were﻿5,407﻿and﻿5,402﻿respectively﻿for﻿the﻿rank﻿A﻿group,﻿
3,553﻿and﻿3,586﻿respectively﻿for﻿the﻿rank﻿B﻿group,﻿and﻿1,849﻿and﻿1,808﻿respectively﻿for﻿the﻿rank﻿C﻿
group.﻿The﻿reading﻿time﻿was﻿categorized﻿into﻿three﻿ranks:﻿A﻿(the﻿top﻿33%),﻿B﻿(the﻿middle﻿33%)﻿and﻿
C﻿(the﻿bottom﻿34%).﻿The﻿students﻿were﻿classified﻿into﻿these﻿three﻿types﻿accordingly.﻿Table﻿3﻿shows﻿
the﻿parts﻿of﻿transaction﻿data.﻿Based﻿on﻿Felder﻿and﻿Soloman﻿(2001),﻿the﻿four﻿dimensions﻿of﻿LSI﻿were﻿
divided﻿into﻿LSI1﻿(Active﻿or﻿Reflective),﻿LSI2﻿(Sensing﻿or﻿Intuitive),﻿LSI3﻿(Visual﻿or﻿Verbal),﻿and﻿
LSI4﻿(Sequential﻿or﻿Global).﻿Based﻿on﻿Allinson﻿et﻿al.﻿(1996),﻿the﻿ranges﻿of﻿CSI﻿scores﻿were﻿classified﻿
into﻿three﻿cognitive﻿styles,﻿namely﻿Analytic,﻿Adaptive,﻿and﻿Intuitive﻿((Table﻿2).

The﻿analysis﻿detected﻿5623﻿association﻿rules﻿(Figure﻿3).﻿The﻿horizontal﻿axis﻿represents﻿the﻿support﻿
value,﻿and﻿the﻿vertical﻿axis﻿represents﻿the﻿confidence﻿value.﻿Support﻿is﻿an﻿indication﻿of﻿how﻿frequently﻿
the﻿detected﻿rules﻿appear﻿in﻿the﻿database;﻿thus,﻿support﻿is﻿the﻿relative﻿frequency﻿of﻿transactions﻿that﻿
contain﻿X﻿and﻿Y﻿(X﻿and﻿Y﻿are﻿item﻿sets).﻿Confidence﻿is﻿an﻿indication﻿of﻿how﻿often﻿the﻿rule﻿has﻿been﻿
found﻿to﻿be﻿true.﻿This﻿study﻿decided﻿the﻿two﻿regions﻿(1)﻿and﻿(2)﻿through﻿expert﻿human﻿judgment﻿with﻿
the﻿detected﻿association﻿rules.

Region﻿(1)﻿includes﻿the﻿association﻿rules﻿whose﻿support﻿is﻿less﻿than﻿0.1,﻿and﻿its﻿confidence﻿is﻿less﻿
than﻿0.6.﻿The﻿experts﻿were﻿not﻿able﻿to﻿find﻿important﻿association﻿rules﻿such﻿as﻿the﻿relations﻿among﻿
learning﻿styles,﻿digital﻿textbook﻿logs,﻿and﻿learning﻿achievements﻿if﻿the﻿support﻿value﻿is﻿less﻿than﻿0.1﻿and﻿
the﻿confidence﻿value﻿is﻿less﻿than﻿0.6.﻿Therefore,﻿we﻿do﻿not﻿consider﻿the﻿association﻿rules﻿of﻿region﻿(1).

Table 1. Sample action log

Userid Action Name Document ID Page Number Action Time

Student1 Next 00000000NBU4 16 2014/10/22﻿8:40:55

Student1 Prev 00000000NBU4 15 2014/10/22﻿8:42:15

Student2 Add﻿Marker 00000000NBU4 15 2014/10/22﻿8:42:16

Student3 Add﻿Memo 00000000NBU4 15 2014/10/22﻿8:42:18

Table 2. The ranking of reading time

Rank Criteria Sum of reading time 
(seconds)

Mean 
(seconds)

Median 
(seconds)

A Top﻿33% 91,919 5,407 5,042

B Middle﻿33% 53,303 3,553 3,586

C Bottom﻿34% 29,590 1,849 1,808


International Journal of Distance Education Technologies
Volume 17 • Issue 1 • January-March 2019

7

Region﻿(2)﻿includes﻿the﻿association﻿rules﻿whose﻿support﻿is﻿greater﻿than﻿0.1,﻿and﻿its﻿confidence﻿is﻿
greater﻿than﻿0.6.﻿The﻿number﻿of﻿detected﻿association﻿rules﻿in﻿this﻿region﻿was﻿712.﻿This﻿study﻿analyzed﻿
the﻿relationships﻿among﻿the﻿test﻿scores,﻿reading﻿times,﻿LSI,﻿and﻿CSI﻿based﻿on﻿the﻿association﻿rules﻿
with﻿detected﻿high﻿support﻿and﻿confidence﻿values.

RESULT

First,﻿to﻿determine﻿the﻿relationships﻿among﻿CSI﻿and﻿other﻿parameters,﻿we﻿detected﻿association﻿rules﻿
if﻿the﻿Right-Hand-Side﻿(RHS)﻿represents﻿the﻿Analytic﻿type﻿of﻿CSI.﻿Table﻿4﻿shows﻿the﻿association﻿
rules﻿containing﻿one﻿or﻿two﻿factors﻿in﻿the﻿Left-Hand-Side﻿(LHS)﻿and﻿RHS﻿with﻿the﻿Analytic﻿type﻿of﻿
CSI.﻿The﻿color﻿of﻿each﻿cell﻿represents﻿the﻿confidence﻿value﻿if﻿the﻿support﻿value﻿is﻿greater﻿than﻿0.1.﻿
For﻿example,﻿(1)﻿in﻿Table﻿4﻿indicates﻿one﻿association﻿rule﻿if﻿LHS﻿is﻿“LSI1﻿=﻿Reflective”﻿and﻿“LSI2﻿
=﻿Sensing”﻿and﻿RHS﻿is﻿“CSI﻿=﻿Analytic,”﻿and﻿the﻿confidence﻿value﻿is﻿0.626.

Table 3. The parts of transaction data

ID OP Pre-test Post-test Reading time LSI1 LSI2 LSI3 LSI4 CSI

1 NEXT 80 90 A Active Sensing Visual Sequential Analytic

2 NEXT 80 90 A Active Sensing Verbal Sequential Adaptive

3 PREV 70 80 B Reflective Intuitive Visual Global Intuitive

4 NEXT 80 90 A Reflective Intuitive Verbal Global Intuitive

Figure 3. The distribution of the detected association rules


International Journal of Distance Education Technologies
Volume 17 • Issue 1 • January-March 2019

8

When﻿comparing﻿each﻿learning﻿style﻿through﻿the﻿pre-test﻿and﻿post-test﻿sores,﻿association﻿rules﻿
were﻿found﻿when﻿the﻿students﻿had﻿a﻿reflective﻿type﻿of﻿LSI,﻿and﻿the﻿pre-test﻿score﻿was﻿80,﻿post-test﻿
score﻿90,﻿and﻿post-test﻿score﻿100.﻿These﻿association﻿rules﻿had﻿significant﻿relationships﻿because﻿their﻿
confidence﻿levels﻿are﻿higher﻿than﻿other﻿factors﻿such﻿as﻿the﻿sensing﻿or﻿visual﻿sequential﻿type﻿of﻿LSI.﻿
When﻿comparing﻿each﻿learning﻿style﻿with﻿the﻿reading﻿time,﻿association﻿rules﻿were﻿found﻿in﻿the﻿cases﻿
where﻿the﻿students﻿have﻿the﻿reflective﻿type﻿of﻿LSI﻿and﻿“reading﻿time﻿=﻿A.”﻿This﻿fact﻿indicates﻿that﻿
students﻿of﻿the﻿reflective﻿type﻿with﻿“reading﻿time﻿=﻿A”﻿fit﻿into﻿the﻿analytic﻿type﻿of﻿CSI.

Following﻿the﻿above﻿results,﻿this﻿study﻿further﻿explored﻿the﻿data﻿to﻿find﻿association﻿rules﻿in﻿cases﻿
where﻿the﻿RHS﻿represents﻿the﻿reflective﻿or﻿active﻿types﻿of﻿LSI.﻿Table﻿5﻿shows﻿the﻿cross-tabulation﻿
of﻿LSI﻿with﻿the﻿reflective﻿type.﻿The﻿association﻿rules﻿between﻿each﻿learning﻿style,﻿such﻿as﻿sensing,﻿
visual﻿ sequential,﻿ and﻿ global,﻿ and﻿ other﻿ factors﻿ were﻿ found,﻿ but﻿ these﻿ association﻿ rules﻿ had﻿ lower﻿
confidence﻿values.

Significant﻿association﻿rules﻿among﻿“reading﻿time﻿=﻿A”﻿and﻿pre-test﻿and﻿post-test﻿scores﻿were﻿
found.﻿For﻿the﻿association﻿rule﻿between﻿pre-test﻿score﻿80﻿and﻿post-test﻿score﻿90,﻿the﻿confidence﻿value﻿
was﻿1.﻿There﻿were﻿seven﻿students﻿who﻿increased﻿their﻿test﻿scores﻿by﻿10﻿points﻿from﻿pre-test﻿80﻿to﻿
post-test﻿90.﻿They﻿fit﻿into﻿the﻿reflective﻿type﻿of﻿LSI.﻿Table﻿6﻿shows﻿the﻿cross-tabulation﻿of﻿the﻿LSI﻿
with﻿the﻿active﻿type.﻿For﻿the﻿association﻿rule﻿between﻿pre-test﻿90﻿and﻿post-test﻿80,﻿there﻿were﻿six﻿
students﻿who﻿decreased﻿their﻿test﻿scores﻿by﻿10﻿points﻿from﻿pre-test﻿90﻿to﻿post-test﻿80.﻿They﻿fit﻿into﻿
the﻿active﻿type﻿of﻿LSI.﻿Among﻿the﻿association﻿rules﻿of﻿reading﻿time﻿were﻿association﻿rules﻿indicating﻿
that﻿LHS﻿is﻿“reading﻿time﻿B”﻿or﻿“reading﻿time﻿C.”

Table 4. Cross-tabulation of the CSI with analytics


International Journal of Distance Education Technologies
Volume 17 • Issue 1 • January-March 2019

9

Table﻿7﻿shows﻿the﻿cross-tabulation﻿of﻿LSI﻿with﻿the﻿visual﻿type.﻿Association﻿rules﻿were﻿found﻿
indicating﻿ that﻿ LHS﻿ was﻿ associated﻿ with﻿ reflective,﻿ sensing,﻿ global,﻿ and﻿ sequential﻿ types﻿ of﻿ LSI;﻿
however,﻿these﻿association﻿rules﻿had﻿a﻿lower﻿confidence﻿value,﻿that﻿is,﻿less﻿than﻿0.8.

Among﻿the﻿association﻿rules﻿of﻿CSI,﻿it﻿was﻿found﻿that﻿the﻿confidence﻿value﻿of﻿the﻿adaptive﻿type﻿
was﻿higher﻿than﻿that﻿of﻿the﻿analytic﻿type,﻿while﻿among﻿the﻿association﻿rules﻿of﻿reading﻿time,﻿there﻿
were﻿association﻿rules﻿whereby﻿LHS﻿fell﻿into﻿“reading﻿time﻿A,”﻿“reading﻿time﻿B,”﻿or﻿“reading﻿time﻿
C.”﻿When﻿comparing﻿each﻿reading﻿time﻿with﻿high﻿confidence﻿values,﻿association﻿rules﻿were﻿found﻿
such﻿that﻿LHS﻿fell﻿under﻿the﻿reflective﻿type﻿of﻿LSI﻿with﻿“reading﻿time﻿A”﻿and﻿that﻿LHS﻿fell﻿under﻿the﻿
sensing﻿or﻿global﻿type﻿of﻿LSI﻿with﻿“reading﻿time﻿C.”﻿This﻿means﻿that﻿a﻿majority﻿of﻿the﻿students﻿with﻿
the﻿reflective﻿type﻿of﻿LSI﻿with﻿“reading﻿time﻿A”﻿or﻿sensing﻿or﻿global﻿type﻿of﻿LSI﻿with﻿“reading﻿time﻿
C”﻿were﻿associated﻿with﻿the﻿visual﻿type﻿of﻿LSI.﻿Among﻿the﻿association﻿rules﻿concerning﻿the﻿pre-test﻿
and﻿post-test﻿scores,﻿a﻿meaningful﻿association﻿rule﻿was﻿found﻿that﻿LHS﻿fell﻿under﻿pre-test﻿score﻿90﻿
with﻿post-test﻿score﻿90.﻿This﻿means﻿that﻿students﻿who﻿received﻿a﻿pre-test﻿score﻿90﻿and﻿a﻿post-test﻿
score﻿of﻿90﻿or﻿100﻿fit﻿into﻿the﻿visual﻿type﻿of﻿LSI.

From﻿these﻿meaningful﻿association﻿rules,﻿this﻿study﻿mainly﻿categorized﻿three﻿student﻿types:﻿Poorly﻿
Motivated,﻿Efficient,﻿and﻿Diligent﻿(Figure﻿4).﻿The﻿diligent﻿type﻿meets﻿five﻿conditions:﻿Pre-test﻿score﻿
80,﻿Post-test﻿score﻿90,﻿“LSI﻿=﻿Reflective,”﻿“CSI﻿=﻿Analytic,”﻿and﻿“Reading﻿time﻿A.”﻿The﻿efficient﻿type﻿
fulfills﻿five﻿conditions:﻿Pretest﻿score﻿90﻿or﻿100,﻿Post-test﻿score﻿90,﻿“LSI﻿=﻿Visual,”﻿“CSI﻿=﻿Adaptive,”﻿
and﻿“Reading﻿time﻿=﻿A,﻿B,﻿or﻿C.”﻿The﻿poorly﻿motivated﻿type﻿satisfies﻿four﻿conditions:﻿Pre-test﻿score﻿
90,﻿Post-test﻿score﻿80,﻿“reading﻿time﻿B﻿or﻿C,”﻿and﻿“LSI﻿=﻿Active.”﻿Based﻿on﻿these﻿conditions﻿as﻿well﻿
as﻿the﻿questionnaire﻿results﻿for﻿LSI﻿and﻿CSI,﻿the﻿three﻿student﻿types﻿were﻿identified.

Table 5. Cross-tabulation of the LSI with the reflective type


International Journal of Distance Education Technologies
Volume 17 • Issue 1 • January-March 2019

10

DISCUSSIoN

From﻿ the﻿ results﻿ in﻿ Figure﻿ 4,﻿ this﻿ study﻿ identified﻿ three﻿ student﻿ types.﻿ Poorly﻿ motivated﻿ type﻿ is﻿
characterized﻿by﻿a﻿lower﻿reading﻿time﻿of﻿digital﻿textbooks﻿than﻿other﻿types.﻿We﻿considered﻿that﻿the﻿
active﻿students﻿did﻿not﻿prefer﻿thinking﻿about﻿and﻿reflecting﻿on﻿things﻿through﻿reading﻿digital﻿textbooks.﻿
Previous﻿studies﻿have﻿reported﻿that﻿active﻿learners﻿learn﻿by﻿doing﻿something﻿with﻿the﻿information﻿
obtained.﻿They﻿prefer﻿to﻿process﻿information﻿by﻿talking﻿about﻿it﻿and﻿trying﻿it﻿out.﻿Therefore,﻿when﻿
teachers﻿identify﻿the﻿student﻿type﻿in﻿the﻿questionnaire﻿stage,﻿it﻿is﻿necessary﻿to﻿design﻿interactive﻿digital﻿
textbooks﻿that﻿students﻿can﻿interact﻿with﻿functions﻿such﻿as﻿audios,﻿videos,﻿and﻿so﻿on.

The﻿efficient﻿type﻿of﻿students﻿includes﻿the﻿adaptive﻿type﻿of﻿CSI﻿and﻿the﻿visual﻿type﻿of﻿LSI.﻿The﻿
adaptive﻿style﻿implies﻿a﻿balanced﻿blend﻿of﻿intuition﻿and﻿analytical﻿style.﻿In﻿addition,﻿visual﻿learners﻿
prefer﻿ visual﻿ presentations﻿ of﻿ materials.﻿ They﻿ like﻿ pictures,﻿ diagrams,﻿ graphs,﻿ and﻿ charts.﻿ In﻿ the﻿
evaluation﻿experiments,﻿students﻿were﻿able﻿to﻿obtain﻿good﻿scores﻿in﻿the﻿pre-test﻿and﻿post-test﻿even﻿
though﻿we﻿were﻿not﻿able﻿to﻿prepare﻿rich﻿digital﻿textbooks﻿with﻿the﻿ideal﻿number﻿of﻿pictures,﻿diagrams,﻿
and﻿graphs.﻿Thus,﻿our﻿next﻿analysis﻿should﻿be﻿more﻿carefully﻿planned﻿(with﻿well-designed﻿digital﻿
textbooks﻿or﻿not).

On﻿the﻿other﻿hand,﻿a﻿diligent﻿type﻿has﻿a﻿higher﻿reading﻿time﻿of﻿digital﻿textbooks﻿than﻿other﻿types.﻿
Reflective﻿ learners﻿ learn﻿ by﻿ thinking﻿ about﻿ information.﻿ They﻿ prefer﻿ to﻿ think﻿ things﻿ through﻿ and﻿
understand﻿things﻿before﻿acting.﻿Therefore,﻿it﻿is﻿necessary﻿to﻿design﻿digital﻿textbooks﻿that﻿promote﻿
critical﻿thinking.

CoNCLUSIoN

This﻿ study﻿ analyzed﻿ the﻿ learning﻿ patterns﻿ or﻿ rules﻿ using﻿ digital﻿ textbook﻿ logs﻿ with﻿ LSI﻿ and﻿ CSI﻿
questionnaires.﻿To﻿collect﻿students’﻿digital﻿textbook﻿logs,﻿the﻿DITel﻿system﻿was﻿developed﻿and﻿used﻿
in﻿ a﻿ commercial﻿ law﻿ course﻿ for﻿ undergraduate﻿ students.﻿ The﻿ data﻿ collection﻿ period﻿ for﻿ the﻿ digital﻿
textbook﻿logs﻿was﻿from﻿March﻿to﻿July﻿2017.﻿A﻿total﻿of﻿41﻿undergraduate﻿students﻿participated﻿in﻿this﻿
study.﻿They﻿were﻿asked﻿to﻿answer﻿the﻿LSI﻿and﻿CSI﻿questionnaires﻿to﻿investigate﻿their﻿learning﻿and﻿
cognitive﻿styles.

Table 6. Cross-tabulation of the LSI with the active type


International Journal of Distance Education Technologies
Volume 17 • Issue 1 • January-March 2019

11

Table 7. Cross-tabulation of the LSI with the visual type

Figure 4. Three student types categorized based on the detected association rules


International Journal of Distance Education Technologies
Volume 17 • Issue 1 • January-March 2019

12

This﻿study﻿used﻿association﻿analysis﻿with﻿the﻿Apriori﻿algorithm.﻿Using﻿the﻿analysis﻿method,﻿we﻿
found﻿5,265﻿association﻿rules.﻿Furthermore,﻿we﻿checked﻿to﻿find﻿meaningful﻿association﻿rules﻿using﻿
human﻿judgment﻿in﻿accordance﻿with﻿the﻿indications﻿of﻿support﻿and﻿confidence.

Based﻿ on﻿ the﻿ association﻿ rules,﻿ we﻿ found﻿ the﻿ following﻿ categories:﻿ The﻿ criteria﻿ of﻿ “CSI﻿ =﻿
Analytic”﻿and﻿“LSI﻿=﻿Reflective”﻿on﻿RHS﻿and﻿meaningful﻿conditions﻿detected﻿of﻿“Reading﻿time﻿A,”﻿
“Pre-test﻿score﻿80,”﻿and﻿“Post-test﻿score﻿90”﻿categorize﻿the﻿diligent﻿type.﻿“LSI﻿=﻿Visual”﻿and﻿“CSI﻿
=﻿Adaptive”﻿on﻿RHS﻿and﻿meaningful﻿conditions﻿detected﻿of﻿“Reading﻿time﻿A,﻿B,﻿or﻿C,”﻿“Pre-test﻿
score﻿90﻿or﻿100,”﻿and﻿“Post-test﻿score﻿90”﻿categorize﻿the﻿efficient﻿type.﻿Finally,﻿“LSI﻿=﻿Active”﻿on﻿
RHS﻿and﻿meaningful﻿conditions﻿detected﻿such﻿as﻿“Reading﻿time﻿B﻿or﻿C,”﻿“Pre-test﻿score﻿90,”﻿and﻿
“Post-test﻿score﻿80”﻿categorize﻿the﻿poorly﻿motivated﻿type.

In﻿sum,﻿the﻿main﻿contribution﻿of﻿this﻿study﻿is﻿to﻿find﻿learning﻿patterns﻿or﻿rules﻿for﻿enhancing﻿
education.﻿By﻿considering﻿the﻿detected﻿learning﻿patterns,﻿we﻿believe﻿that﻿teachers﻿can﻿provide﻿support﻿
particularly﻿to﻿students﻿of﻿the﻿poorly﻿motivated﻿type﻿in﻿advance.﻿In﻿the﻿future,﻿we﻿will﻿consider﻿a﻿
dashboard﻿development﻿(Lkhagvasuren﻿et﻿al.,﻿2016)﻿to﻿predict﻿student﻿types﻿in﻿accordance﻿with﻿their﻿
learning﻿logs﻿and﻿LSI﻿and﻿CSI﻿questionnaires.

ACKNowLEDGMENT

Part﻿of﻿this﻿research﻿work﻿was﻿also﻿supported﻿by﻿the﻿Grant-in-Aid﻿for﻿Scientific﻿Research﻿No.16H03078﻿
and﻿No.﻿17K12947﻿from﻿the﻿Ministry﻿of﻿Education,﻿Culture,﻿Sports,﻿Science﻿and﻿Technology﻿(MEXT)﻿
in﻿Japan.


International Journal of Distance Education Technologies
Volume 17 • Issue 1 • January-March 2019

13

REFERENCES

Allinson,﻿C.﻿W.,﻿&﻿Hayes,﻿J.﻿(1996).﻿The﻿cognitive﻿style﻿index:﻿A﻿measure﻿of﻿intuition-analysis﻿for﻿organizational﻿
research.﻿Journal of Management Studies,﻿33(1),﻿119–136.﻿doi:10.1111/j.1467-6486.1996.tb00801.x

Baker,﻿R.﻿S.,﻿&﻿Inventado,﻿P.﻿S.﻿(2014).﻿Educational﻿data﻿mining﻿and﻿learning﻿analytics.﻿In﻿Learning analytics﻿
(pp.﻿61–75).﻿New﻿York,﻿NY:﻿Springer.

Behrouz,﻿M.,﻿Gerd,﻿K.,﻿&﻿William,﻿P.﻿(2004).﻿Association﻿analysis﻿for﻿an﻿online﻿education﻿system.﻿In﻿Proceedings 
of the 2004 IEEE international conference on Information Reuse and Integration﻿(pp.﻿504-509).﻿Las﻿Vegas,﻿NV:﻿
Institute﻿of﻿Electrical﻿and﻿Electronics﻿Engineers.

Felder,﻿R.﻿M.,﻿&﻿Soloman,﻿B.﻿A.﻿(2001).﻿Learning﻿styles﻿and﻿strategies.﻿Retrieved﻿from﻿http://www.engr.ncsu.
edu/learningstyles/ilsweb.html

Greller,﻿ W.,﻿ &﻿ Drachsler,﻿ H.﻿ (2012).﻿ Translating﻿ learning﻿ into﻿ numbers:﻿ A﻿ generic﻿ framework﻿ for﻿ learning﻿
analytics.﻿Journal of Educational Technology & Society,﻿15(3),﻿42–57.

Gu,﻿X.,﻿Wu,﻿B.,﻿&﻿Xu,﻿X.﻿(2015).﻿Design,﻿development,﻿and﻿learning﻿in﻿e-Textbooks:﻿What﻿we﻿learned﻿and﻿where﻿
we﻿are﻿going.﻿Journal of Computers in Education,﻿2(1),﻿25–41.﻿doi:10.1007/s40692-014-0023-9

Hezroni,﻿O.﻿E.﻿(2004).﻿Literacy﻿and﻿assistive﻿technology﻿for﻿children﻿with﻿special﻿needs.﻿Script,﻿7-8,﻿195–218.﻿
(in﻿Hebrew)

Hwang,﻿G.﻿J.,﻿Chu,﻿H.﻿C.,﻿&﻿Yin,﻿C.﻿(2017).﻿Objectives,﻿methodologies﻿and﻿research﻿issues﻿of﻿learning﻿analytics.﻿
Interactive Learning Environments,﻿25(2),﻿143–146.﻿doi:10.1080/10494820.2017.1287338

Hwang,﻿G.﻿J.,﻿Hsu,﻿T.﻿C.,﻿Lai,﻿C.﻿L.,﻿&﻿Hsueh,﻿C.﻿J.﻿(2017).﻿Interaction﻿of﻿problem-based﻿gaming﻿and﻿learning﻿
anxiety﻿ in﻿ EFL﻿ students’﻿ English﻿ listening﻿ performance﻿ and﻿ progressive﻿ behavioral﻿ patterns.﻿ Computers & 
Education,﻿106,﻿26–42.﻿doi:10.1016/j.compedu.2016.11.010

Lee,﻿R.,﻿Zickuhr,﻿K.,﻿Purcell,﻿K.,﻿Madden,﻿M.,﻿&﻿Brenner,﻿J.﻿(2012).﻿The rise of e-reading.﻿Washington,﻿D.C:﻿
Pew﻿Research﻿Center’s﻿Internet﻿&﻿American﻿Life﻿Project.

Lkhagvasuren,﻿E.,﻿Matsuura,﻿K.,﻿Mouri,﻿K.,﻿&﻿Ogata,﻿H.﻿(2016).﻿Dashboard﻿for﻿analyzing﻿ubiquitous﻿learning﻿
log.﻿International Journal of Distance Education Technologies,﻿14(3),﻿1–20.﻿doi:10.4018/IJDET.2016070101

Mouri,﻿K.,﻿Ogata,﻿H.,﻿&﻿Uosaki,﻿N.﻿(2015).﻿Ubiquitous﻿learning﻿analytics﻿in﻿the﻿context﻿of﻿real-world﻿language﻿
learning.﻿In﻿Proceedings of the Learning Analytics and Knowledge﻿(pp.﻿378-382).﻿Poughkeepsie,﻿NY:﻿Association﻿
for﻿Computing﻿Machinery.﻿doi:10.1145/2723576.2723598

Mouri,﻿ K.,﻿ Ogata,﻿ H.,﻿ &﻿ Uosaki,﻿ N.﻿ (2016).﻿ Context-aware﻿ and﻿ personalization﻿ method﻿ based﻿ on﻿ ubiquitous﻿
learning﻿analytics.﻿Journal of Universal Computer Science,﻿22(10),﻿1380–1397.

Mouri,﻿K.,﻿Okubo,﻿F.,﻿Shimada,﻿A.,﻿&﻿Ogata,﻿H.﻿(2016).﻿Bayesian﻿network﻿for﻿predicting﻿students’﻿final﻿grade﻿
using﻿e-book﻿logs﻿in﻿university﻿education.﻿In﻿2016 IEEE 16th International Conference on Advanced Learning 
Technologies (ICALT)﻿(pp.﻿85-89).﻿Texas,﻿TX:﻿Institute﻿of﻿Electrical﻿and﻿Electronics﻿Engineers.

Mouri,﻿K.,﻿&﻿Yin,﻿C.﻿(2017).﻿E-book-based﻿learning﻿analytics﻿for﻿improving﻿learning﻿materials.﻿In﻿T.﻿Matsuo﻿
et﻿ al.﻿ (Eds),﻿ Proceedings of the 6th IIAI International Congress on Advanced Applied Informatics﻿ (pp.﻿ 493-
497).﻿Hamamatsu,﻿Japan:﻿International﻿Institute﻿of﻿Applied﻿Informatics.﻿Institute﻿of﻿Electrical﻿and﻿Electronics﻿
Engineers.﻿doi:10.1109/IIAI-AAI.2017.59

Ogata,﻿ H.,﻿ Hou,﻿ B.,﻿ Uosaki,﻿ N.,﻿ Mouri,﻿ K.,﻿ &﻿ Liu,﻿ S.﻿ (2014).﻿ Ubiquitous﻿ learning﻿ project﻿ using﻿ life-logging﻿
technology﻿in﻿Japan.﻿Journal of Educational Technology & Society,﻿17(2),﻿85–100.

Package﻿arule.﻿(2017).﻿Mining﻿Association﻿Rules﻿and﻿Frequent﻿Itemsets.﻿Retrieved﻿from﻿https://cran.r-project.
org/web/packages/arules/arules.pdf

Reinking,﻿D.﻿(1997).﻿Me﻿and﻿my﻿hypertext:﻿A﻿multiple﻿digression﻿analysis﻿of﻿technology﻿and﻿literacy﻿(sic).﻿The 
Reading Teacher,﻿50,﻿626–643.

Rockinson-Szapkiw,﻿A.﻿J.,﻿Courduff,﻿J.,﻿Carter,﻿K.,﻿&﻿Bennett,﻿D.﻿(2013).﻿Electronic﻿versus﻿traditional﻿print﻿
textbooks:﻿A﻿comparison﻿study﻿on﻿the﻿influence﻿of﻿university﻿students’﻿learning.﻿Computers & Education,﻿63,﻿
259–266.﻿doi:10.1016/j.compedu.2012.11.022

http://dx.doi.org/10.1111/j.1467-6486.1996.tb00801.x
http://www.engr.ncsu.edu/learningstyles/ilsweb.html
http://www.engr.ncsu.edu/learningstyles/ilsweb.html
http://dx.doi.org/10.1007/s40692-014-0023-9
http://dx.doi.org/10.1080/10494820.2017.1287338
http://dx.doi.org/10.1016/j.compedu.2016.11.010
http://dx.doi.org/10.4018/IJDET.2016070101
http://dx.doi.org/10.1145/2723576.2723598
http://dx.doi.org/10.1109/IIAI-AAI.2017.59
https://cran.r-project.org/web/packages/arules/arules.pdf
https://cran.r-project.org/web/packages/arules/arules.pdf
http://dx.doi.org/10.1016/j.compedu.2012.11.022


International Journal of Distance Education Technologies
Volume 17 • Issue 1 • January-March 2019

14

Kousuke Mouri is an assistant professor at the institute of Engineering, Tokyo University of Agriculture and 
Technology, Japan. His research interests include computer supported ubiquitous and mobile learning, augmented 
reality, data mining and network analysis for authentic learning. He received the PhD degree at the Graduate School 
of Information Science and Electrical Engineering, Kyushu University. He received best student paper award in 
the ICCE 2014 international conference. He is a member of SOLAR (Society of Learning Analytics and Research), 
and APSCE (Asian Pacific Society for Computer in Education).

Zhuo Ren is a lecturer in international school of Ji Nan university, China. She received her LL.M. in 2005 from 
Law School of Sun Yat-Sen university, China. She is focused on the study of legal teaching and practising. She 
is also a practicing lawyer in China.

Noriko Uosaki is currently an associated professor at the Center for International Education and Exchange, Osaka 
University, Osaka, Japan. She received the Ph.D. degree in educational technology from Tokushima University 
in 2013. Her research interests include MALL (Mobile Assisted Language Learning), Seamless Learning, CALL 
(Computer Assisted Language Learning), Computer Supported Ubiquitous and Mobile Learning, CSCL (Computer 
Supported Collaborative Learning), and TESL (Teaching English as a Second Language). She is a member of 
JSET, IEEE, and APSCE.

Chengjiu Yin is an Associate Professor at the Information Science and Technology Center, Kobe University, Japan. 
He received his PhD degree from the Department of Information Science and Intelligent Systems, Tokushima 
University, Japan, in 2008. Currently he is committing himself in mobile learning, ubiquitous computing, language 
learning, and educational data mining. He is a member of JSET, JSiSE and APSCE.

Shepperd,﻿J.﻿A.,﻿Grace,﻿J.﻿L.,﻿&﻿Koch,﻿E.﻿J.﻿(2008).﻿Evaluating﻿the﻿electronic﻿textbook:﻿Is﻿it﻿time﻿to﻿dispense﻿with﻿
the﻿paper﻿text?﻿Teaching of Psychology,﻿35(1),﻿2–5.﻿doi:10.1177/009862830803500102

Shimada,﻿A.,﻿Okubo,﻿F.,﻿Yin,﻿C.,﻿&﻿Ogata,﻿H.﻿(2017).﻿Automatic﻿summarization﻿of﻿lecture﻿slides﻿for﻿enhanced﻿
student﻿preview:﻿Technical﻿report﻿and﻿user﻿study.﻿IEEE Transactions on Learning Technologies,﻿99.﻿doi:10.1109/
TLT.2017.2682086

Snyder,﻿I.﻿(2002).﻿Silicon literacies: Communication, innovation and education in the electronic age.﻿London:﻿
Routledge.

Yin,﻿C.,﻿Okubo,﻿F.,﻿Shimada,﻿A.,﻿Kentaro,﻿K.,﻿Yamada,﻿M.,﻿Ogata,﻿H.,﻿&﻿Fujimura,﻿N.﻿(2014).﻿Smart﻿phone﻿
based﻿data﻿collecting﻿system﻿for﻿analyzing﻿learning﻿behaviors.﻿In﻿Proceedings of the International Conference 
of Computers on Education,﻿Nov.﻿30﻿-﻿Dec.﻿4﻿(pp.﻿575-577).﻿Nara,﻿Japan:﻿Asia-Pacific﻿Society﻿for﻿Computers﻿
in﻿Education.

Yin,﻿C.,﻿Okubo,﻿F.,﻿Shimada,﻿A.,﻿Oi,﻿M.,﻿Hirokawa,﻿S.,﻿Yamada,﻿M.,﻿.﻿.﻿.﻿Ogata,﻿H.﻿(2015).﻿Analyzing﻿the﻿features﻿
of﻿learning﻿behaviors﻿of﻿students﻿using﻿e-books.﻿In﻿Doctoral Student Consortium (DSC) -﻿Proceedings of the 
23rd International Conference on Computers in Education, ICCE 2015﻿(pp.﻿617-626).﻿Hangzhou,﻿China:﻿Asia-
Pacific﻿Society﻿for﻿Computers﻿in﻿Education.

Yin,﻿C.﻿J.,﻿Hirokawa,﻿S.,﻿Yau,﻿J.,﻿Nakatoh,﻿T.,﻿Hashimoto,﻿K.,﻿&﻿Tabata,﻿Y.﻿(2013).﻿Analyzing﻿research﻿trends﻿
with﻿cross﻿tabulation﻿search﻿engine.﻿International Journal of Distance Education Technologies,﻿11(1),﻿31–44.﻿
doi:10.4018/jdet.2013010103

Yin,﻿C.﻿J.,﻿Sung,﻿H.﻿Y.,﻿Hwang,﻿G.﻿J.,﻿Hirokawa,﻿S.,﻿Chu,﻿H.﻿C.,﻿Flanagan,﻿B.,﻿&﻿Tabata,﻿Y.﻿(2013).﻿Learning﻿by﻿
searching:﻿A﻿learning﻿approach﻿that﻿provides﻿searching﻿and﻿analysis﻿facilities﻿to﻿support﻿research﻿trend﻿surveys.﻿
Journal of Educational Technology & Society,﻿16(3),﻿286–300.

Yin,﻿ C.﻿ J.,﻿ Uosaki,﻿ N.,﻿ Chu,﻿ H.﻿ C.,﻿ Hwang,﻿ G.﻿ J.,﻿ Hwang,﻿ J.﻿ J.,﻿ Hatono,﻿ I.,﻿ .﻿ .﻿ .﻿ Tabata,﻿ Y.﻿ (2017).﻿ Learning﻿
behavioral﻿pattern﻿analysis﻿based﻿on﻿students’﻿logs﻿in﻿reading﻿digital﻿books.﻿In﻿W.﻿Chen﻿et﻿al.﻿(Eds),﻿Proceedings 
of the 25th International Conference on Computers in Education﻿(pp.﻿549-557).﻿Christchurch,﻿New﻿Zealand:﻿
Asia-Pacific﻿Society﻿for﻿Computers﻿in﻿Education.

http://dx.doi.org/10.1177/009862830803500102
http://dx.doi.org/10.1109/TLT.2017.2682086
http://dx.doi.org/10.1109/TLT.2017.2682086
http://dx.doi.org/10.4018/jdet.2013010103