7
Lehrstuhl Informatik 5 (Information Systems) Prof. Dr. M. Jarke 1 Layers These slides are licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License. A Dynamic Topic Model of Learning Analytics Research Past – Present – Future Michael Derntl*, Nikou Günnemann and Ralf Klamma RWTH Aachen University Advanced Community Information Systems (ACIS) Informatik 5, Aachen, Germany * [email protected] LAK Data Challange 2013 April 9, 2013 Leuven, Belgium

A Dynamic Topic Model of Learning Analytics Research

Embed Size (px)

DESCRIPTION

Presented at LAK Data Challenge, Linked Data Tutorial, LAK 2013 Conference April 9, Leuven, Belgium

Citation preview

Page 1: A Dynamic Topic Model of Learning Analytics Research

Lehrstuhl Informatik 5

(Information Systems)

Prof. Dr. M. Jarke

1

Layers

These slides are licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.

A Dynamic Topic Model of

Learning Analytics Research

Past – Present – Future

Michael Derntl*, Nikou Günnemann and Ralf Klamma

RWTH Aachen University Advanced Community Information Systems (ACIS)

Informatik 5, Aachen, Germany

* [email protected]

LAK Data Challange 2013

April 9, 2013

Leuven, Belgium

Page 2: A Dynamic Topic Model of Learning Analytics Research

Lehrstuhl Informatik 5

(Information Systems)

Prof. Dr. M. Jarke

2

Layers

Methodology: Dynamic Topic

Modeling

Relational DB

(ID, venue, year,

title, authors,

abstract, fulltext,

hyperlink)

RDF XML

Dump

(LAK + EDM)

RDF XML to

Relational DB

Paper 1 text blah

blah blah blah

Paper N text blah

blah blah blah

Probabilistic

topic mining

(LDA)

Figure © ACM. Source: D. M. Blei: Probabilistic topic models. Commun. ACM 55(4): 77-84 (2012)

Dynamic topic

modeling of LAK

data (20 topics,

5 epochs)

Page 3: A Dynamic Topic Model of Learning Analytics Research

Lehrstuhl Informatik 5

(Information Systems)

Prof. Dr. M. Jarke

3

Layers

Main Topics in LAK Dataset

"students model parameters skill" (A): most relevant and most volatile

"model data features prediction" (B) – high relevance and high stability the LAK topic?

– in 2012 "predicition" most relevant to this topic

"network community discussion analysis" (R): – quite irrelevant and volatile

– rise in relevance through LAK 2011

General prominence of students, learners, model, data in multiple topics

(bubble size: relevance in 2012)

LAK 2011

Page 4: A Dynamic Topic Model of Learning Analytics Research

Lehrstuhl Informatik 5

(Information Systems)

Prof. Dr. M. Jarke

4

Layers

Impact of LAK Conference

Overall topic "river":

Turbulence in EDM era

More shifts through LAK 2011

Catharsis in 2012? EDM '09 EDM '08 EDM '10 EDM '11

LAK '11

EDM '12

LAK '12

ET&S SI

EDM '11

LAK '11

EDM '10

Zooming into 2010-11 transition:

Three topics reach absolute peak

(black up triangles)

– model students data probability

– network community discussion analysis

– problem students model types

Increased focus on network,

community, student models

Page 5: A Dynamic Topic Model of Learning Analytics Research

Lehrstuhl Informatik 5

(Information Systems)

Prof. Dr. M. Jarke

5

Layers

Most Recent Topic Trends

Highest absolute rise in relevance in 2012: 1. students data courses system

2. students interaction participants analysis

3. learning analytics social learners

4. students actions learning state

5. data user learning dataset

From 11% cumulative relevance in 2009 to 42% in 2012

Page 6: A Dynamic Topic Model of Learning Analytics Research

Lehrstuhl Informatik 5

(Information Systems)

Prof. Dr. M. Jarke

6

Layers

Summary

Past

Turbulent topic currents in EDM-only era

LAK and EDM have a shared topic foundation (data modeling, classification, clustering, …)

Present

Emphasis on learner modeling, data modeling, analysis, prediction.

1st LAK conference brought topic shifts (e.g. networks, community)

Future

Topic shifts 2012 moderate; convergence of LAK research topics

Social and interaction aspects; students as research subjects

Page 7: A Dynamic Topic Model of Learning Analytics Research

Lehrstuhl Informatik 5

(Information Systems)

Prof. Dr. M. Jarke

7

Layers

Do It Yourself:

Topic Analytics with D-VITA

http://is.gd/laktopics