41
Mo#va#on Data on the Web Some eyecatching opener illustra#ng growth and or diversity of web data LAK – Data Challenge 2014 #LAKdata14 Stefan Dietze , Eelco Herder (L3S Research Center, DE) Mathieu d’Aquin (The Open University, UK) Davide Taibi (Ins=tute for Educa=onal Technologies CNR, IT) Hendrik Drachsler (Welten Ins=tute, Open Universiteit Nederland, NL) 09/04/13 Hendrik Drachsler

LAK14 Data Challenge

Embed Size (px)

DESCRIPTION

What do analytics on learning analytics tell us? How can we make sense of this emerging field’s historical roots, current state, and future trends, based on how its members report and debate their research? Challenge submissions should exploit the LAK Dataset for a meaningful purpose. This may include submissions which cover one or more of the following, non-exclusive list of topics: Analysis & assessment of the emerging LAK community in terms of topics, people, citations or connections with other fields Innovative applications to explore, navigate and visualise the dataset (and/or its correlation with other datasets) Usage of the dataset as part of recommender systems Analysis of the evolution of LAK discipline Improvement or enrichment of the LAK Dataset

Citation preview

Page 1: LAK14 Data Challenge

Mo#va#on  Data  on  the  Web  

Some  eyecatching  opener  illustra#ng  growth  and  or  diversity  of  web  data      

LAK  –  Data  Challenge  2014    #LAKdata14  

Stefan  Dietze  ,  Eelco  Herder  (L3S  Research  Center,  DE)  Mathieu  d’Aquin  (The  Open  University,  UK)  

Davide  Taibi  (Ins=tute  for  Educa=onal  Technologies  CNR,  IT)  Hendrik  Drachsler  (Welten  Ins=tute,  Open  Universiteit  Nederland,  NL)  

09/04/13 Hendrik Drachsler

Page 2: LAK14 Data Challenge

26 March 2014 Hendrik Drachsler 2

Page 3: LAK14 Data Challenge

3

• Assistant  Professor  on  Technology-­‐Enhanced  Learning  

• Research  topics:  Personaliza=on,    Recommender  Systems,    Learning  Analy=cs,    Mobile  devices  

• Applica=on  domains:    Science  2.0  Health  2.0  

WhoAmI

Page 4: LAK14 Data Challenge

4

Research Projects

Page 5: LAK14 Data Challenge

09/04/13

LAK  Challenge  –  sponsored  by  LinkedUp  

§  EC-­‐funded  support  ac#on,  started  in  11/2012  

§  Three  pillars:  

§  LinkedUp  Challenge:  open  data  compe==on  (over  1.5  years)    [  hVp://linkedup-­‐challenge.org  ]  

§  Linked  Educa#on  Data:  data  &  catalog  for  large-­‐scale  educa=onal  Web  data  applica=ons  

§  Evalua#on  Framework  for  open  data  applica=ons  

 

 

http://www.linkedup-project.eu/

Partners  

Stefan Dietze

Page 6: LAK14 Data Challenge

LinkedUp Consortium

Page 7: LAK14 Data Challenge

LinkedUp Network

Page 8: LAK14 Data Challenge

LinkedUp  Data  Catalogue    

hVp://data.linkededuca=on.org/linkedup/catalog/  

Page 9: LAK14 Data Challenge

LinkedUp  Data  Catalog  in  a  nutshell  

Stefan Dietze 26 March 2014

hVp://data.linkededuca=on.org/linkedup/categories-­‐explorer  

§  RDF  dataset  catalog  of  Linked  Open  Data  for  learning  

§  Browse,  explore  and  query  across  the  LOD  cloud  

§  Federated  queries  

Page 10: LAK14 Data Challenge
Page 11: LAK14 Data Challenge

§  Open  &  focused  track(s)  

§  Final  events  at  ESWC2014    (May,  Crete)  

§  Open  Track  only          

§  Final  events  at  OKCon  2013  (September  2013,  Geneva)  

§  Open  track  &  focused  tracks    

 §  Submission  details  and  calls  to  be  

released  soon  

§  Final  events  at  ISWC2014    (October,  Riva  del  Garda,  Italy)  

May  –September  2013   October  2013  –  May  2014   May  2014  –  October  2014  

?  

Page 12: LAK14 Data Challenge

Veni  shortlist  &  winners  DataConf.    

KnowNodes    

Mismuseos    

ReCredible  

YourHistory    

26 March 2014

hVp://www.globe-­‐town.org/  

WeShare  -­‐  3rd  price  /  people‘s  choice  

GlobeTown  -­‐  2nd  price  

hVp://seek.cloud.gsic.tel.uva.es/weshare/  

hVp://www.polimedia.nl/  

PoliMedia  –  1st    price  

Page 13: LAK14 Data Challenge

1st  Place:  PoliMedia  Exploring  poli=cal  debates  &  events  

09/04/13 Stefan Dietze

§  Explora#on  of  poli#cal  debates  and  news  coverage  

§  Automa=cally  generated  links  between  transcripts  debates,  newspaper  ar#cles,  

§  Generated  data  available  as  Linked  Data  (hVp://data.polimedia.nl)  

§  Data  sources:  1)  newspapers  in  their  original  layout  of  the  historical  newspaper  archive,  and  2)  radio  bulle=ns  of  the  Dutch  Na=onal  Press  Agency  (ANP)  

§  9000+  debates  (1945  –  1995)  

§  Over  3000  media  links  

Mar=jn  Kleppe,  Max  Kemman,  Henri  Beunders  (Erasmus  Universiteit  RoVerdam),  Laura  Hollink  Damir  Juric  (Vrije  Universiteit  Amsterdam),  Johan  Oomen  Jaap  Blom  (Nederlands  Ins=tuut  voor  Beeld  en  Geluid)  

hVp://www.polimedia.nl/  

Page 14: LAK14 Data Challenge

2nd  Place:  GlobeTown  Open  data  for  sustainable  development  

09/04/13 Stefan Dietze

§  Exploring  interac=on  between  data  about  environment,  economy  and  society  (cause  and  effect  in  complex  systems)  

§  Diverse  open  government  data:    

§  SWERA  renewable  energy  data:  hVp://swera.unep.net/,    

§  UN  Comtrade  trade  data:  hVp://comtrade.un.org/,    

§  Country  names  and  codes:  hVp://opengeocode.org,    

§  OECD  sta=s=cs:  hVp://stats.oecd.org/,  

§  World  Bank:  hVp://data.worldbank.org    

§  Top  three  at  Apps4Climate  compe==on,  held  by  the  World  Bank  

Jack  Townsend,  Andrea  Prieto-­‐Vega,  Richard  Gomer,  Will  Fyson,  Dom  Hobson,  Huw  Fryer  (University  of  Southampton,  UK)  

hVp://www.globe-­‐town.org/  

Page 15: LAK14 Data Challenge

3rd  Place/Peoples  Choice:  WeShare  Exploring,  annota=ng,  ra=ng  educa=onal  ICT  tools  

Stefan Dietze 26 March 2014

§  Social  &  seman=c  annota#on  applica=on  for  educa#onal  ICT  tools  

§  Aids  educators  to  find  tools  to  support  teaching  at  all  educa=onal  levels  

§  Gathers  and  enriches  data  from  exis=ng  registries  and  datasets  

§  Currently:  approx.  7000  tool  descrip#ons  

§  Crowdsourcing:  educators,  tutors  etc  can  modify  and  enrich  data    

Adolfo  Ruiz-­‐Calleja,  Guillermo  Vega-­‐Gorgojo,  Juan  I.  Asensio-­‐Pérez,  Eduardo  Gómez-­‐Sánchez,  Miguel  L.  Bote-­‐Lorenzo,  Carlos  Alario-­‐Hoyos  Universidad  de  Valladolid,  Universidad  Carlos  III  de  Madrid,  Valladolid,  Spain  

hVp://seek.cloud.gsic.tel.uva.es/weshare/  

Page 16: LAK14 Data Challenge

Veni  Awards,  OKCon,  Geneva  

•  Live streaming of awards •  900+ at event •  Video of awards online

Page 17: LAK14 Data Challenge

17

November 2013 – May 2014

Page 18: LAK14 Data Challenge

Vidi  Compe##on    

Focused Tracks

Looking for prototypes that solve a particular problem

Open Track

Page 19: LAK14 Data Challenge

Overview

•  14 entries from 12 countries •  Currently being looked at by review panel •  Assessing using evaluation framework •  Winners will be announced at ESWC in Crete, May 2014 •  Prizes for open track and focused track •  People’s Choice

Vidi  Compe##on    

Page 20: LAK14 Data Challenge

May – October 2013

•  Soon to be launched! •  Awards ceremony at ISWC, Italy, October 2014

Page 21: LAK14 Data Challenge

Learning  Analy#cs  &  Knowledge  Data  &  Challenge    Facilita=ng  Research  on  Learning  Analy=cs  and  EDM  

26 March 2014

hVp://lak.linkededuca=on.org/  

LAK  Dataset  (450  publica#ons  in  RDF/R)  §  ACM  Interna=onal  Conference  on  Learning  Analy=cs  and  

Knowledge  (LAK)  (2011-­‐13)  §  Interna=onal  Conference  on  Educa=onal  Data  Mining  (2008-­‐13)  §  Journal  of  Educa=onal  Data  Mining  (2008-­‐12)  

LAK  Data  Challenge  §  Analyse,  explore  correlate  the  LAK  Dataset  §  At  ACM  LAK  2014  (April  2014,  Indianapolis)  

Page 22: LAK14 Data Challenge

Learning  Analy#cs  &  Knowledge  (LAK)  Dataset  

§  A  corpus  of  metadata  and  full-­‐text  of    all  learning  analy=cs  &  educa=onal  data  mining  publica=ons    

§  Freely,  openly  available  in  variety  of  structured  formats  

§  Open  access  as  well  as  previously  non-­‐public  resources  

 

Publica#on   #  of  papers  

Proceedings  of  the  ACM  Interna#onal  Conference  on  Learning  Analy#cs  and  Knowledge  (LAK)  (2011-­‐12)  

66  

The  open  access  journal  Educa#onal  Technology  &  Society  special  issue  on   “Learning   and   Knowledge   Analy#cs”:   Educa=onal   Technology   &  Society   (Special   Issue   on   Learning   &   Knowledge   Analy=cs,   edited   by  George  Siemens  &  Dragan  Gašević),  2012,  15,  (3),  pp.  1-­‐163.  

10  

Proceedings   of   the   Interna#onal   Conference   on   Educa#onal   Data  Mining  (2008-­‐12)  

239  

Journal  of  Educa#onal  Data  Mining  (2008-­‐12)   16  

Special  permission    from  ACM  

26 March 2014 22 Hendrik Drachsler

Page 23: LAK14 Data Challenge

Learning  Analy#cs  &  Knowledge  (LAK)  Dataset  Extrac=on  process  

   

26 March 2014

Further reading: Taibi, D., Dietze, S., Fostering Analytics on Learning Analytics Research: the LAK Dataset, in: CEUR WS Proceedings Vol. 974, Proceedings of the LAK Data Challenge, held at LAK2013 – The Third Conference on Learning Analytics and Knowledge, April 2013., URL: http://ceur-ws.org/Vol-974/lakdatachallenge2013_preface.pdf Hendrik Drachsler

Page 24: LAK14 Data Challenge

LAK  Challenge  Submissions  (accepted)in  a  nutshell  

4%

7%4%

17%

3%

3%

21%

17%

14%

10%

authors

Brazil

Canada

France

Germany

Italy

Netherlands

Serbia

Spain

United  Kingdom

26 March 2014 24 Hendrik Drachsler

Page 25: LAK14 Data Challenge

LAK  Challenge  Proceedings,  CEUR  Vol-­‐974  a  nutshell  

http://ceur-ws.org/Vol-974

Page 26: LAK14 Data Challenge

#LAKdata13–  the  many  faces  of  a  small  dataset  

Analysis Exploration & Visualisation

Search & Recommendation Correlation & Enrichment

Page 27: LAK14 Data Challenge

#LAKdata13  –  Most  used  Technologies  2013  

1.  NLP  tools  (Stanford  Parser  &  parts-­‐of-­‐speech  POS  tagger),  Sta=s=cal  &  graph  based  techniques,  Concept  Filtering  (JUNG  framework),  Degree  centrality,  Betweeness  centrality,  HITS  algorithm  [importance  of  hubs  and  authori=es],  PageRank,  TF-­‐IDF  

2.  NLP  (concept  extrac=on  +  rhetorical  analysis),  Xerox  Incremental  Parser  rhetorical  analysis,  RapidMiner,  yED  tool  (clusters  of  docs),  PHP,  Javascript,  Google  chart  tools  

3.  MS  Excel,  NodeXL,  NetDraw,  IBM  Many  Eyes    LD  Source  (DPSpotlight  Alchemy  API)  4.  TF-­‐IDF,  Mahout  system,  Wekin  tool,  SNA  5.  Dbpedia,  Geonames,  DBLp-­‐Gnoss,  DBPedia,  Geonames,  Enrichment  of  data  6.  Programming  (D3,  PHP,  MySQL),  TagMe,  Dbpedia  Spotlight,  TF-­‐IDF,  Wikipedia  Miner  7.  Dbpedia  Spotlight,  Dbpedia  Knowledge  graph,  t-­‐idf,  Freebase  8.  Jensen-­‐Shannon  divergence  measure,    9.  Tools  (Tableu,  Excel,)  

Create  a  tag  cloud  with  techniques  used  

26 March 2014 LinkedUp – Dr. Hendrik Drachsler 27

Page 28: LAK14 Data Challenge

#LAKdata13  –  Authors  (mostly  EDM)  

•  3.  table  3,  4    •  4.  table  1  degree  centrality  •  7  Figure  1  co-­‐author  network    

26 March 2014 28 Hendrik Drachsler

Page 29: LAK14 Data Challenge

#LAKdata13  –  Authors  (purely  LAK)  

•  3.  table  3,  4    •  4.  table  1  degree  centrality  •  7  Figure  1  co-­‐author  network    

26 March 2014 29 Hendrik Drachsler

Page 30: LAK14 Data Challenge

#LAKdata13  –  Central  countries  

•  3.  Table  5,  Figure  1,2,  table  6,  7  •  5.  figure  2,  4    

26 March 2014 30 Hendrik Drachsler

Page 31: LAK14 Data Challenge

#LAKdata13  –  Central  topics  

•  Learners-­‐model-­‐data    Student-­‐model-­‐parameter-­‐skill  •  Model-­‐data-­‐features-­‐predic=on  •  network-­‐community-­‐discussion-­‐analysis  (2011  )  

26 March 2014 31 Hendrik Drachsler

EDM LAK •  Intelligent  tutoring  systems  •  Accuracy  of  different  types  of  

predic=ve  models  •  Revealing  unexpected  paVerns  •  Feature  extrac=on  •  Types  of  parameter  •  automa=c  support  for  edu.  

Processes  •  Adapta=on  and  personaliza=on  of  

learning  •  predic=on,  accuracy  and  precision  

•  More  focused  on  Teacher  Support  with  LA  to  support  students  

•  Promo=ng  reflec=on  for  students  and  instructors  

•  Informal  learning    •  Leverage  human  judgment  on  

informing  and  empowering  instructors  

•  Empowering  learners  to  reflect  over  learning  processes    

•  Network-­‐community-­‐social-­‐users  

Shared topics

Page 32: LAK14 Data Challenge

#LAKdata13  –  EDM  vs.  LAK  

26 March 2014 32 Hendrik Drachsler

Page 33: LAK14 Data Challenge

#LAKdata13  –  EDM  &  LAK  

1.  Students,  data,  models  2.  Figure  1  +  Clusters  of  whole  dataset  3.  26  single  authors  ar=cles  by  25  authors  Stephan  E.  Fancsali  =  2  single  authored  publica=ons,  SS  

Table  1,  73.5%  of  all  ar=cles  collaborated  only  once,  Table  3  top  collaborators,  Table  4,  5:  top  interna=onal  collaborators    

4.  6.  data  analysis,  2011  1st  Lak  learning  analy=cs  became  popular  topic,    

26 March 2014 33 Hendrik Drachsler

| 2011 – 1st LAK

Page 34: LAK14 Data Challenge

26 March 2014 34

#LAKdata14  –  LAK  Data  Challenge  2014  

Hendrik Drachsler

Page 35: LAK14 Data Challenge

#LAKdata14  –  Agenda  

26 March 2014 LinkedUp – Dr. Hendrik Drachsler 35

Chance for some awesome gadgets:

Page 36: LAK14 Data Challenge

Mo#va#on  Data  on  the  Web  

Some  eyecatching  opener  illustra#ng  growth  and  or  diversity  of  web  data      

LAK  –  Data  Challenge  2015    #LAKdata15  

-­‐=  FOCUS  TASKS  =-­‐  Stefan  Dietze  ,  Eelco  Herder  (L3S  Research  Center,  DE)  

Mathieu  d’Aquin  (The  Open  University,  UK)  Davide  Taibi  (Ins=tute  for  Educa=onal  Technologies  CNR,  IT)  

Hendrik  Drachsler  (Welten  Ins=tute,  Open  Universiteit  Nederland,  NL)  

09/04/13 Hendrik Drachsler

Page 37: LAK14 Data Challenge

LAK  Challenge  –  and  the  winners  are...  

09/04/13

Page 38: LAK14 Data Challenge

LAK  Challenge  –  and  the  winners  are...  

09/04/13

Page 39: LAK14 Data Challenge

LAK  Challenge  –  and  the  winners  are...  

09/04/13

Page 40: LAK14 Data Challenge

LAK  Challenge  –  and  the  winners  are...  

09/04/13

Page 41: LAK14 Data Challenge

41

This silde is available at:http://www.slideshare.com/Drachsler

Email: [email protected]: celstec-hendrik.drachslerBlogging at: http://www.drachsler.deTwittering at: http://twitter.com/HDrachsler

Many thanks for your attention!

26 March 2014 LinkedUp – Dr. Hendrik Drachsler 41