PLE Recommendations
-
Upload
ple-conference-2011-southampton -
Category
Education
-
view
424 -
download
0
Transcript of PLE Recommendations
© role-project.eu
May I suggest? Three PLE recommender strategies in comparison
PLE Conference, Southampton, July 11-13, 2011
Felix Mödritscher (speaker), Barbara KrumayVienna Univ. of Economics &Business, Austria
Sten Govaerts, Erik DuvalKatholieke Universiteit Leuven, Belgium
Ingo DahnUniversity of Koblenz-Landau, Germany
Sandy El Helou, Denis GilletEPFL, Switzerland
Alexander Nussbaumer, Dietrich Albert Graz University of Technology, Austria
Carsten UllrichShanghai Jiao Tong University, China
11-13/07/2011 PLE Conference,Southampton, 2011 2/9 , © role-project.eu
Agenda
1. PLEs and recommendations – why?
2. Three approaches from the ROLE project
a. Federated Search Widget
b. Community-based PLE Recommender
c. Psycho-pedagogical Recommender
3. Comparison of the PLE recommenders
1. Personal Learning Environments (PLEs)
PLE = “set of tools, services, and artefactsgathered from various contexts andto be used by learners” [Henri et al., 08]
Characteristics of PLE-based activities: several actors (different roles) ... ... use technology (tools) to ... ... connect to learner networks and ... ... collaborate on shared artefacts ... ... in order to achieve common goals. [Wild, 09]
Problems: Learners/teachers have varying attitudes and skills in using ICT! can cause negative feelings or states (frustration, distraction, etc.); hindering to proceed with learning or failing to achieve goals[Windschitl & Sahl, 02; Nguyen-Ngoc & Law, 08]
11-13/07/2011 PLE Conference,Southampton, 2011 3/9 , © role-project.eu
1. PLEs and Recommendations
Recommendations are necessary “if users have to make choices without sufficient personal experiences of alternatives” [Resnick & Varian, 97]
For TEL: Examples described in the RecSysTEL workshop proceedings!
For PLEs:
1. Pre-given PLE designs for specific needs
2. Possible PLE entities (artefacts, tools, peers) helpful for a specific situation
Recommendations are a powerful instrument for empowering learners to design their PLEs and use technology for learning...
However: Different solution approaches driven by different disciplines...
And: CF techniques not sufficient! (global vs. local top-n)
11-13/07/2011 PLE Conference,Southampton, 2011 4/9 , © role-project.eu
2. Approach 1: Federated search widget ‘Binocs’
Aggregate heterogeneous resources from different (social media) repositories
Save, share, assess, and repurpose resources according to user’s interests
Actions taken into account: select resource, like/dislike, preview
Learning/social context derived from course
Forward contextual data to a recommender system (3A contextual ranking service, Graaasp [El Helou et al., 09])
Ranking according to previous interactions and relevance to search query
11-13/07/2011 PLE Conference,Southampton, 2011 5/9 , © role-project.eu
http://widgetstore.role-demo.de/content/binocs
2. Approach 2: Community-based recommender ‘PLEShare’
Practice sharing repository on the Web; to be integrated into PLE solutions (Web-API)
Idea: users share PLE experiences voluntarily Two demos: (a) PLEShare widget, (b) PAcMan add-on PAcMan: allows designing tool bundles in the form of
tagged bookmarking lists (=activities); simple features for sharing such activities and retrieving/reusing them
Shared data used for generatingtwo kinds of recommendations:(1) activity patterns for starting newactivities [‘Pattern Store’](2) top-n PLE items (artefacts,tools, peers) for a specific context[no explicit feature but availablevia Web-API]
Techniques: CF, clustering
11-13/07/2011 PLE Conference,Southampton, 2011 6/9 , © role-project.eu
https://addons.mozilla.org/en-US/firefox/addon/176479
http://teldev.wu.ac.at/pleshare/api/
2. Approach 3: Psycho-pedagogical recommender
Developed according to theoretical models (self-regulated learning) and relevant taxonomies [Fruhmann et al., 10]
Based on learning goals and competences (learner monitoring and questionnaires)
Realised as widget for providing:(1) support for planning new activities;(2) guidance for ongoing activities; to findappropriate resources (artifacts, tools, peers)
Additional features planned: allowing learners to give feedback on recommendations (implicitly through usage data); provision of explanations; visual feedback on planned and completed activities
Techniques: rule/model-based recommender Remark: no full-featured prototype available
11-13/07/2011 PLE Conference,Southampton, 2011 7/9 , © role-project.eu
http://widgetstore.role-demo.de/content/navigation-tool-widget
3. Comparison of our PLE recommenders
11-13/07/2011
PLE Conference,Southampton, 2011 8/9 , © role-project.eu
Binocs widget PLEShare PP recommender
recommender strategy
CF, PageRank-like & content-based
CF & IR/clustering (cliques, topics, ...)
rule/profile-based (competences)
data & data gathering
on entering search terms, automated
tagged bookmarks, voluntarily shared
questionnaires, automated (profile)
estimated accuracy
high (works well in specialized scope; fallback through IR)
average (requires ‘initialization’, cf. cold start & sparsity)
average (rules and profile must be given)
PLE scenario support & usability
average (PLE design phase not considered)
good (currently only focus on PLE design); usable prototypes
good; restricted to pre-def. domains; no cold-start problem
privacy concerns
sufficient anonymization
privacy statement, anonymized activity recordings (=patterns)
raw usage data not used; user profiles not addressed yet
preliminary experiences
preferences for Google results; uptake in business setting better
three studies; works but requires pilot users sharing patterns (e.g. teachers)
internal evaluations; efforts to integrate new data; requires modelling expertise
Please vote for our mediacast
if you like the idea of PLE practice sharing!
http://vimeo.com/groups/ple2011/videos/25817690
11-13/07/2011 PLE Conference,Southampton, 2011 9/9 , © role-project.eu
Thanks for your attention!