Data analytics to support awareness and recommendation
1 like935 views
This document discusses data analytics and recommender systems. It provides an overview of research topics related to learning analytics applications and healthcare analytics. Recommender system algorithms like user-based and item-based collaborative filtering are examined. Challenges with contextual recommendation systems are outlined, like standardizing contextual data representation and evaluation. Visualization techniques for recommender systems are proposed to make results more comprehensible to users and allow them to provide feedback to steer the information mining process.
Data analytics to support awareness and recommendation
- 1. Data analytics to support awareness and recommendation Katrien Verbert WISE research group Department of Computer Science katrien.verbert@vub.ac.be 27/03/14
- 2. Data analytics Src: Steve Schoettler
- 6. Student Activity Meter (SAM) 6 Govaerts, S., Verbert, K., Duval, E., & Pardo, A. (2012, May). The student activity meter for awareness and self-reflection. In CHI'12 EA (pp. 869-884). ACM.
- 8. Design Based Research Methodology ¤ Rapid prototyping ¤ Evaluate Ideas in short iteration cycles of Design, Implementation & Evaluation ¤ Focus on Usefulness & Usability ¤ Think-aloud evaluations, SUS (System Usability Scale) surveys, usability lab, ...
- 9. demographics tool deployed tracking tools data tracked #cgiar 19 teachers SAM LMS resource use, communication, time spent #lak11 12 participants SAM LMS resource use, communication, time spent #uc3m 11 teachers SAM Virtual machine resource use, programming errors, debugging, time spent; artefacts produced #thesis11 13 students Step-Up! Twitter, Tinyarm, blogs resource use, artefacts produced #thesis11- sup 5 teachers Step-Up! Twitter, Tinyarm, blogs resource use, artefacts produced #peno3 10 students Step-Up! Toggl Time spent, resource and application use #chikul 30 students Step-Up! Toggl, twitter, blogs twitter, blogs, time spent, resource use
- 10. Evaluation results 10 Verbert, K., Govaerts, S., Duval, E., Santos, J. L., Van Assche, F., Parra, G., & Klerkx, J. Learning dashboards: an overview and future research opportunities. Personal and Ubiquitous Computing, 1-16. http://link.springer.com/article/10.1007/s00779-013-0751-2
- 15. similarity measures ¤ Cosine similarity ¤ Pearson correlation ¤ Tanimoto or extended Jaccard coefficient
- 16. similarity measures 16 MAE of item-based collaborative filtering based on different similarity metrics
- 17. algorithms MAE of user-based, item-based and slope-one collaborative filtering
- 18. data dimensions
- 21. Challenges ¤ context acquisition ¤ standardized representation of contextual data ¤ evaluation ¤ user interfaces
- 23. Problem statement ¤ Complexity prevents users from comprehending results ¤ Trust issues when recommendations fail ¤ Aggravated with contextual recommendation ¤ The black box nature of RS prevents users from providing feedback ¤ Algorithms typically hard-wired in the system code ¤ generate a list of top-N recommendations ¤ little research has been done to study more flexible approaches 23
- 25. Interrelations agents – users - tags 25
- 26. Interrelations agents – users 26
- 27. Interrelations agents - tags 27
- 28. TalkExplorer 28
- 29. effectiveness How frequently a specific combination type produced a display that was used to bookmark at least one interesting item Dimensions of relevance are not equal The more aspects of relevance are used, the more effective it is Especially effective are fusions across relevance dimensions 29
- 31. 31 information visualisation - information retrieval - information (data) mining
- 33. Combining information mining and visualization Core objectives: • make mining results comprehensible for users • enable users to steer the information mining process