SlideShare a Scribd company logo
Mixed-initiative Recommender Systems
Katrien Verbert
Augment/HCI - KU Leuven
@katrien_v
Human-Computer Interaction group
PhD. researcher Oct. 2003 – Feb. 2008
Post-doc Feb. 2008 – Dec. 2012
Assistant Professor Jan. 2013 – Sept 2014
Assistant Professor Oct. 2014 – Sept. 2018
learning analytics – recommender systems – visualisation
Associate Professor Oct. 2018 – …
Augment/HCI team
Robin De Croon
Postdoc researcher
Katrien Verbert
Associate Professor
Francisco Gutiérrez
PhD researcher
Tom Broos
PhD researcher
Martijn Millecamp
PhD researcher
Sven Charleer
Postdoc researcher
Nyi Nyi Htun
Postdoc researcher
Houda Lamqaddam
PhD researcher
Yucheng Jin
PhD researcher
Oscar Alvarado
PhD researcher
http://augment.cs.kuleuven.be/
Diego Rojo Carcia
PhD researcher
Peter Brusliovsky Nava Tintarev Cristina ConatiDenis Parra
Collaborations
7
Verbert, Katrien; Manouselis, Nikos; Ochoa, Xavier;Wolpers, Martin; Drachsler, Hendrik; Bosnic, Ivana; Duval, Erik. Context-
aware recommender systems for learning: a survey and future challenges, IEEE Trans. on Learning Technologies, 18 p. (2012)
8
9
Combining recommendation and visualization
Core objectives:
• make recommendations understandable for users
• enable users to steer the recommendation process
Flexible interaction with RecSys
Research visit
¤ Host: Carnegie Mellon
University & University of
Pittsburg
¤ Collaboration: John Stamper,
Peter Brusilovsky, Denis Parra
¤ Period: April 2012 – June 2012
(3 months)
Second post-doctoral
fellowship FWO
¤ host university: KU Leuven,
Belgium
¤ supervisor: Erik Duval
¤ period: Oct 2012 – Sept 2015
11
Overview research topics
12
2012 | 2013 | 2014 | 2015 | 2016 | 2017 | 2018
Learning Analytics - Media Consumption – Research Information Systems - Healthcare
Motivation
1
3
¤ multiple relevance prospects in
personalized social tagging
systems
¤ community relevance prospects
¤ social relevance prospect
¤ content relevance prospect
¤ existing personalized social
systems
¤ do not allow to explore and combine
multiple relevance prospects
¤ only one prospect can be explored at a
given time
Also recommendations è personalized relevance
prospect
14
Shortcomings recommender systems
15
¤ cold-start issues
¤ difficult to explain the rationale behind recommendations
¤ user control often limited
16
He, C., Parra, D., & Verbert, K. (2016). Interactive recommender systems: A survey of the
state of the art and future research challenges and opportunities. Expert Systems with
Applications, 56, 9-27.
John O'Donovan, Barry Smyth, Brynjar Gretarsson, Svetlin Bostandjiev, and Tobias Höllerer. 2008. PeerChooser: visual
interactive recommendation. CHI '08
Related work: PeerChooser
Smallworlds
18
Gretarsson, B., O'Donovan, J., Bostandjiev, S.,
Hall, C. and Höllerer, T. SmallWorlds: Visualizing
Social Recommendations. Comput. Graph. Forum,
29, 3 (2010), 833-842.
Related work: TasteWeights
19
Bostandjiev,S.,O'Donovan,J.andHöllerer,T.TasteWeights:avisualinteractive
hybridrecommendersystem.InProceedingsofthesixthACMconferenceon
Recommendersystems(RecSys'12).ACM,NewYork,NY,USA(2012),35-42.
Contributions
¤ new approach to support exploration, transparency
and controllability
¤ recommender systems are shown as agents
¤ in parallel to real users and tags
¤ users can interrelate entities to find items
¤ evaluation study that assesses
¤ effectiveness
¤ probability of item selection
20
Verbert, K., Parra, D., Brusilovsky, P., & Duval, E. (2013). Visualizing recommendations
to support exploration, transparency and controllability. In Proceedings of the IUI
2013 international conference on Intelligent user interfaces (pp. 351-362). ACM.
Conference Navigator
The studies were conducted using Conference Navigator 3
http://halley.exp.sis.pitt.edu/cn3/
TalkExplorer - I
Entities
Tags, Recommender
Agents, Users
22
TalkExplorer - II
23
Our Assumptions
¤ Items which are relevant in more that one
aspect could be more valuable to the users
¤ Displaying multiple aspects of relevance visually
is important for the users in the process of item’s
exploration
24
User study 1
¤ Setup
¤ supervised user study
¤ 21 participants at UMAP 2012 and ACM Hypertext 2012
conferences
¤ Procedure
¤ Tasks
¤ interact with users and their bookmarks
¤ interact with agents
¤ interact with tags
¤ Post-questionnaire
25
Evaluation
¤ Data collection
¤ recordings of voice and screen using camtasia studio
¤ system logs
¤ Measurements
¤ effectiveness: # bookmarked items / #explorations
¤ yield: : # bookmarked items / sum of items in selection
set
26
Summary results
Sign. effect. Sign.
yield
multiple versus one entity 0.003 <0.001
user versus (user + entity) 0.593 <0.001
agent versus (agent + entity) 0.341 <0.001
User study 2
¤ Setup
¤ Unsupervised user study
¤ Conducted at LAK 2012 and ECTEL 2013 (18 users)
¤ Subjects familiar with visualizations, but not much with
RecSys
¤ Procedure
¤ Users were left free to explore the interface.
¤ Interactions were logged
¤ Post-questionnaire
28
Verbert, K., Parra, D., & Brusilovsky, P. (2016). Agents vs. users: Visual recommendation of research
talks with multiple dimension of relevance. ACM Transactions on Interactive Intelligent Systems
(TIIS), 6(2), 11.
Summary results
29
Sign. effect.
multiple versus one entity <0.001
user versus (user + entity) 0.3682
agent versus (agent + entity) 0.4426
Results of Studies 1 & 2
¤ Effectiveness increases with
intersections of more
entities
¤ Effectiveness wasn’t
affected in the field study
(study 2)
¤ … but exploration
distribution was affected
30
Average effectiveness
Total number of explorations
Drawback: visualizing intersections
Venn diagram: more natural way to visualize intersections
31
Clustermap Venn diagram
Verbert, K., Parra, D., Brusilovksy, P. (2014). The effect of different set-based visualizations on
user exploration of recommendations. In : IntRS@RecSys, 2014 (pp. 37-44).
IntersectionExplorer (IEx)
32
Cardoso, B., Sedrakyan, G., Gutiérrez, F., Parra, D., Brusilovsky, P., & Verbert, K. (2018). IntersectionExplorer, a
multi-perspective approach for exploring recommendations. International Journal of Human-Computer Studies.
IntersectionExplorer (IEx)
33
Research questions
¤ RQ1: Under which condition may a scalable visualisation
increase user acceptance of recommended items?
¤ RQ2: Does a scalable set visualisation increase perceived
effectiveness of recommendations.
¤ RQ3: Does a scalable set visualisation increase user trust
in recommendations?
¤ RQ4: Does a scalable set visualisation improve user
satisfaction with a recommender system?
34
Study 1
¤ Within-subjects study with 20 users
¤ Mean age: 32.9 years; SD: 6.32; female: 3
¤ Baseline: exploration of recommendations in CN3
¤ Second condition: exploration of recommendations in IE
¤ Data from two conferences:
¤ EC-TEL 2014 (172 items)
¤ EC-TEL 2015 (112 items)
35
Results
36
Subjective feedback
Questionnaire results with statistical significance. Differences between
the aspects “Fun” and “Choice satisfaction” were not significant after
the Bonferroni-Holm correction.
37
Study 2: Digital Humanities
38
¤ 39 users, less technically oriented
¤ Mean age: 38 years; SD: 10; female: 11
¤ Data from DH conference: +1000 participants
Results
39
Study 3: IUI 2017
40
¤ 42 users, technically oriented
¤ Mean age: 32.4 years; SD: 10; female: 17
¤ Data from IUI conference: 111 accepted papers
Results
41
TalkExplorer vs IntersectionExplorer
42
Study 1 vs Study 2 vs Study 3
¤ Overall ”augmented agents” were used in all three
studies
¤ Participants of study 2 (Digital Humanities)
¤ more interested in content perspective
¤ Rated several dimensions lower (use intention, fun,
information sufficiency, control)
43
Overview research topics
44
2012 | 2013 | 2014 | 2015 | 2016 | 2017 | 2018
Learning analytics - Media Consumption – Research Information Systems - Healthcare
Research questions
¤ How do personal characteristics influence the impact of
¤ RQ1: user controls in terms of diversity and acceptance?
¤ RQ2: visualizations in terms of diversity and acceptance?
¤ RQ3: visualizations + user controls in terms of diversity and
acceptance?
45
46
Study procedure
¤ Tutorial of study
¤ Pre-study questionnaire:
¤ Demographics
¤ Experience
¤ Visual Memory (VM) Capacity
¤ Musical Sophistication (MS)
¤ Task: Based on a chosen scenario for creating a play-list, explore songs
and rate all songs in the final playlist
¤ Post-study questionnaire:
¤ perceived quality
¤ perceived accuracy
¤ perceived diversity
¤ Satisfaction
¤ Effectiveness
¤ choice difficulty
Experimental platform
The Spotify API allows to generate recommendations
¤ based on up to five favourite artists.
¤ 14 musical attributes in order to describe musical
preference
48
Dependent variables
¤ Perceived diversity: self-reported measure based on
questionnaire
¤ Recommendation acceptance: measured by
percentage of liked songs in the play-list
49
Personal characteristics
¤ Musical sophistication (MS):
¤ measurement of the ability to engage with music in a
flexible, effective and nuanced way (Müllensiefen et al.,
2014)
¤ Measured using the Goldsmiths Musical Sophistication Index
(Gold-MSI)
¤ Visual memory (VM):
¤ the ability to recall visual patterns (Tintarev and Mastoff,
2016)
¤ Measured by Corsi block-tapping test
50
User-centered factors
¤ Exp.1 and Exp.3 used framework of Knijnenburg et al. 2012
¤ Perceived quality: participants' perceived quality of the
recommended songs.
¤ Perceived accuracy: participants' perceived accuracy of the
recommended songs according to their preference.
¤ Perceived diversity: the similarity among the recommended
songs.
¤ Satisfaction: participants' satisfaction about their chosen
recommendations.
¤ Choice difficulty: difficulty of choosing a recommended song.
¤ Effectiveness: usefulness of recommendations generated from
systems.
¤ Exp. 2 used the ResQue framework
51
Hypotheses
¤ H1: The UI setting (user control, visualization, or both) has a significant
effect on recommendation acceptance.
¤ H2: The UI setting (user control, visualization, or both) has a significant
effect on perceived diversity.
¤ H3: Visual memory (VM) has a significant effect on recommendation
acceptance.
¤ H4: Visual memory (VM) has a significant effect on perceived diversity.
¤ H5: Musical sophistication (MS) has a significant effect on
recommendation acceptance.
¤ H6: Musical sophistication (MS) has a significant effect on perceived
diversity.
52
Participants
53
Experiment 1: Effects of User control
Jin, Y., Tintarev, N., & Verbert, K. (2018, September). Effects of personal characteristics
on music recommender systems with different levels of controllability. In Proceedings
of the 12th ACM Conference on Recommender Systems (pp. 13-21). ACM.
User profile (PRO) Algorithm parameters (PAR) Recommendations (REC)
Evaluation method
¤ Between-subjects
¤ Independent variable: settings of user control
¤ 2x2x2 factorial design
¤ Dependent variables:
¤ Acceptance (ratings)
¤ Cognitive load (NASA-TLX), MS, VM
¤ Framework Knijnenburg et al.
CFA - validity
questions
Quality
Accuracy
Effectiveness
Choice difficulty
Diversity
Trust
Satisfaction
The fit of our SEM
model
X(2, 98) = 257.410
p <.001
RMSEA= 0.083
CFI = 0.980
TLI = 0.968
CFA - validity
questions
Quality
Accuracy
Effectiveness
Choice difficulty
Diversity (low AVE
value)
Trust (low AVE
value)
Satisfaction
(modification
index)
The fit of our SEM
model
X(2, 98) = 257.410
p <.001
RMSEA= 0.083
CFI = 0.980
TLI = 0.968
Conclusion
¤ Main effects: from REC to PRO to PAR → higher cognitive load
¤ Two-way interaction: does not necessarily result in higher
cognitive load. Adding an additional control component to
PAR increases the acceptance. PRO*PAR has less cognitive
load than PRO and PAR
¤ Three-way interaction: it increases acceptance, and does not
lead to higher cognitive load. Increase interaction times and
accuracy
¤ High MS leads to higher quality, and thereby result in higher
acceptance
59
Experiment 2: effects of visualizations
Yucheng Jin, Nava Tintarev, and Katrien Verbert. 2018. Effects of Individual Traits on
Diversity-Aware Music Recommender User Interfaces. In Proceedings of the 26th
Conference on User Modeling, Adaptation and Personalization (UMAP '18). ACM,
New York, NY, USA, 291-299. DOI: https://doi.org/10.1145/3209219.3209225
UI Design
ComBub
61
UI Design - Baseline
SimBub
62
Research Questions
63
¤ How do personal characteristics influence the impact of
¤ RQ1: user controls in terms of diversity and acceptance?
¤ RQ2: visualizations in terms of diversity and acceptance?
¤ RQ3: visualizations + user controls in terms of diversity and
acceptance?
Study design
¤ Within-subjects
¤ Independent variable:
¤ Type of bubble charts
¤ Dependent variables:
¤ Perceived diversity
¤ Overall usability
¤ Support to identify blind-spots
64
Measurements
¤ VM: Corsi block-tapping test
¤ MS: Goldsmiths Musical Sophistication Index
¤ Support to identify blind spots: #explored genres / #available
genres
¤ Perceived diversity: ResQue (item-item diversity, categorial
diversity, novelty, and serendipity)
¤ Usability: ResQue (usefulness, user’s attitude, behavioral intention)
65
Acceptance
66
H1: The UI setting (visualization) has a significant effect on
recommendation acceptance (Cannot accept)
Perceived diversity
67
H2: The UI setting (visualization) has a significant effect on
perceived diversity (Cannot accept)
Mann-Whitney test (U = 2614.00, p = .08)
Personal characteristics
68
ComBub supports the participants with high VM to gain higher
perceived diversity than SimBub. (H4)
ComBub supports the participants with high MS to gain higher
perceived diversity than SimBub. (H6)
Discussion
¤ In general, visualizing the audio features of music has a limited
impact on perceived diversity
¤ Positive correlation between personal characteristics (MS, VM)
and the perceived diversity in ComBub
69
Experiment 3: Effects of Combining
User Control and Visualizations
70
User interface
¤ Full Control + ComBub ¤ Full Control + SimBub
71
Control + ComBub
72
Control + SimBub
73
Research Questions
74
¤ How do personal characteristics influence the impact of
¤ RQ1: user controls in terms of diversity and acceptance?
¤ RQ2: visualizations in terms of diversity and acceptance?
¤ RQ3: visualizations + user controls in terms of diversity and
acceptance?
Acceptance
75
H1: The UI setting has a significant effect on recommendation
acceptance (Cannot accept)
Perceived diversity
76
Full Control vs Control+ComBub (M=5.43, SD=1.07, p = .005)
Full Control vs Control+SimBub (M=5.22, SD=1.20, p = .035)
Personal characteristics
77
Correlation results on acceptance and perceived diversity (* denotes
statistical significance at the 0.05 level, i.e., p value< 0.05)
Discussion
¤ Adding visualization to Full Control seems to increase diversity
significantly.
¤ Control + SimBub
¤ Positive correlation between MS and two dependent variables
(acceptance, perceived diversity)
¤ Control + ComBub
¤ Positive correlation between personal characteristics (MS, VM) and
the perceived diversity in ComBub
¤ MS also has a positive effect on acceptance
¤ => users with high MS are good at tuning the recommender to
find high quality recommendations
78
Overview research topics
79
2012 | 2013 | 2014 | 2015 | 2016 | 2017 | 2018
Learning Analytics - Media Consumption – Research Information Systems - Healthcare
Additional personal characteristics
¤ Need for cognition, visual literacy, …
¤ In collaboration with Cristina Conati
80
Recommender systems for health
81
Augmented reality
82
Gutiérrez, Francisco, Htun, Nyi Nyi, Charleer, Sven, De Croon, Robin, Verbert,
Katrien (2019) Designing Augmented Reality Applications for Personal
Health Decision-Making. Proceedings of HICSS-52.
Tangible Algorithms
¤ Study with Netflix users
¤ Semiotic inspection
¤ Design workshop
¤ Interviews
¤ Abstract representations
¤ Archetype
representations
83
Alvarado, O., Geerts, D. and Verbert, K. Towards Tangible Algorithms: Exploring
Algorithmic Experience with Users’ Profiling Representations. Submitted to CHI 2019.
Job explorer
84
RECOMMENDER
ALGORITHMS
MACHINE
LEARNING
INTERACTIVE DASHBOARDS
SMART ALERTS
RICH CARE PLANS
OPEN IoT
ARCHITECTURE
Questions?
katrien.verbert@cs.kuleuven.be
@katrien_v
Thank you!
http://augment.cs.kuleuven.be/

More Related Content

Similar to Mixed-initiative recommender systems: towards a next generation of recommender systems through user involvement

Interactive Recommender Systems
Interactive Recommender SystemsInteractive Recommender Systems
Interactive Recommender Systems
Katrien Verbert
 
Human-centered AI: how can we support lay users to understand AI?
Human-centered AI: how can we support lay users to understand AI?Human-centered AI: how can we support lay users to understand AI?
Human-centered AI: how can we support lay users to understand AI?
Katrien Verbert
 
Explaining recommendations: design implications and lessons learned
Explaining recommendations: design implications and lessons learnedExplaining recommendations: design implications and lessons learned
Explaining recommendations: design implications and lessons learned
Katrien Verbert
 
Mixed-initiative recommender systems
Mixed-initiative recommender systemsMixed-initiative recommender systems
Mixed-initiative recommender systems
Katrien Verbert
 
Interactive recommender systems: opening up the “black box”
Interactive recommender systems: opening up the “black box”Interactive recommender systems: opening up the “black box”
Interactive recommender systems: opening up the “black box”
Katrien Verbert
 
Human-centered AI: towards the next generation of interactive and adaptive ex...
Human-centered AI: towards the next generation of interactive and adaptive ex...Human-centered AI: towards the next generation of interactive and adaptive ex...
Human-centered AI: towards the next generation of interactive and adaptive ex...
Katrien Verbert
 
Scalable Exploration of Relevance Prospects to Support Decision Making
Scalable Exploration of Relevance Prospects to Support Decision MakingScalable Exploration of Relevance Prospects to Support Decision Making
Scalable Exploration of Relevance Prospects to Support Decision Making
Katrien Verbert
 
Agents vs Users: Visual Recommendation of Research Talks with Multiple Dimens...
Agents vs Users: Visual Recommendation of Research Talks with Multiple Dimens...Agents vs Users: Visual Recommendation of Research Talks with Multiple Dimens...
Agents vs Users: Visual Recommendation of Research Talks with Multiple Dimens...
Katrien Verbert
 
Human-centered AI: how can we support end-users to interact with AI?
Human-centered AI: how can we support end-users to interact with AI?Human-centered AI: how can we support end-users to interact with AI?
Human-centered AI: how can we support end-users to interact with AI?
Katrien Verbert
 
User Control in AIED (Artificial Intelligence in Education)
User Control in AIED (Artificial Intelligence in Education)User Control in AIED (Artificial Intelligence in Education)
User Control in AIED (Artificial Intelligence in Education)
Peter Brusilovsky
 
Preliminary PhD Defence - Student-facing Dashboards
Preliminary PhD Defence - Student-facing DashboardsPreliminary PhD Defence - Student-facing Dashboards
Preliminary PhD Defence - Student-facing Dashboards
Sven Charleer
 
Workshop on Designing Human-Centric MIR Systems
Workshop on Designing Human-Centric MIR SystemsWorkshop on Designing Human-Centric MIR Systems
Workshop on Designing Human-Centric MIR Systems
epsilon_tud
 
Designing Learning Analytics Dashboards: Lessons Learned
Designing Learning Analytics Dashboards: Lessons LearnedDesigning Learning Analytics Dashboards: Lessons Learned
Designing Learning Analytics Dashboards: Lessons Learned
Katrien Verbert
 
Designing and Evaluating Student-facing Learning Dashboards: Lessons Learnt (...
Designing and Evaluating Student-facing Learning Dashboards: Lessons Learnt (...Designing and Evaluating Student-facing Learning Dashboards: Lessons Learnt (...
Designing and Evaluating Student-facing Learning Dashboards: Lessons Learnt (...
Sven Charleer
 
master_thesis.pdf
master_thesis.pdfmaster_thesis.pdf
master_thesis.pdf
EL MAJJODI Ayoub
 
2018-08-23 EARLI Conference in Bonn Quality Reference Framework for MOOCs Str...
2018-08-23 EARLI Conference in Bonn Quality Reference Framework for MOOCs Str...2018-08-23 EARLI Conference in Bonn Quality Reference Framework for MOOCs Str...
2018-08-23 EARLI Conference in Bonn Quality Reference Framework for MOOCs Str...
Christian M. Stracke
 
Interactive recommender systems and dashboards for learning
Interactive recommender systems and dashboards for learningInteractive recommender systems and dashboards for learning
Interactive recommender systems and dashboards for learning
Katrien Verbert
 
2 Studies UX types should know about (Straub UXPA unconference13)
2 Studies UX types should know about (Straub UXPA unconference13)2 Studies UX types should know about (Straub UXPA unconference13)
2 Studies UX types should know about (Straub UXPA unconference13)
Kath Straub
 
Data analytics to support awareness and recommendation
Data analytics to support awareness and recommendationData analytics to support awareness and recommendation
Data analytics to support awareness and recommendationKatrien Verbert
 
Learning to Classify Users in Online Interaction Networks
Learning to Classify Users in Online Interaction NetworksLearning to Classify Users in Online Interaction Networks
Learning to Classify Users in Online Interaction Networks
Symeon Papadopoulos
 

Similar to Mixed-initiative recommender systems: towards a next generation of recommender systems through user involvement (20)

Interactive Recommender Systems
Interactive Recommender SystemsInteractive Recommender Systems
Interactive Recommender Systems
 
Human-centered AI: how can we support lay users to understand AI?
Human-centered AI: how can we support lay users to understand AI?Human-centered AI: how can we support lay users to understand AI?
Human-centered AI: how can we support lay users to understand AI?
 
Explaining recommendations: design implications and lessons learned
Explaining recommendations: design implications and lessons learnedExplaining recommendations: design implications and lessons learned
Explaining recommendations: design implications and lessons learned
 
Mixed-initiative recommender systems
Mixed-initiative recommender systemsMixed-initiative recommender systems
Mixed-initiative recommender systems
 
Interactive recommender systems: opening up the “black box”
Interactive recommender systems: opening up the “black box”Interactive recommender systems: opening up the “black box”
Interactive recommender systems: opening up the “black box”
 
Human-centered AI: towards the next generation of interactive and adaptive ex...
Human-centered AI: towards the next generation of interactive and adaptive ex...Human-centered AI: towards the next generation of interactive and adaptive ex...
Human-centered AI: towards the next generation of interactive and adaptive ex...
 
Scalable Exploration of Relevance Prospects to Support Decision Making
Scalable Exploration of Relevance Prospects to Support Decision MakingScalable Exploration of Relevance Prospects to Support Decision Making
Scalable Exploration of Relevance Prospects to Support Decision Making
 
Agents vs Users: Visual Recommendation of Research Talks with Multiple Dimens...
Agents vs Users: Visual Recommendation of Research Talks with Multiple Dimens...Agents vs Users: Visual Recommendation of Research Talks with Multiple Dimens...
Agents vs Users: Visual Recommendation of Research Talks with Multiple Dimens...
 
Human-centered AI: how can we support end-users to interact with AI?
Human-centered AI: how can we support end-users to interact with AI?Human-centered AI: how can we support end-users to interact with AI?
Human-centered AI: how can we support end-users to interact with AI?
 
User Control in AIED (Artificial Intelligence in Education)
User Control in AIED (Artificial Intelligence in Education)User Control in AIED (Artificial Intelligence in Education)
User Control in AIED (Artificial Intelligence in Education)
 
Preliminary PhD Defence - Student-facing Dashboards
Preliminary PhD Defence - Student-facing DashboardsPreliminary PhD Defence - Student-facing Dashboards
Preliminary PhD Defence - Student-facing Dashboards
 
Workshop on Designing Human-Centric MIR Systems
Workshop on Designing Human-Centric MIR SystemsWorkshop on Designing Human-Centric MIR Systems
Workshop on Designing Human-Centric MIR Systems
 
Designing Learning Analytics Dashboards: Lessons Learned
Designing Learning Analytics Dashboards: Lessons LearnedDesigning Learning Analytics Dashboards: Lessons Learned
Designing Learning Analytics Dashboards: Lessons Learned
 
Designing and Evaluating Student-facing Learning Dashboards: Lessons Learnt (...
Designing and Evaluating Student-facing Learning Dashboards: Lessons Learnt (...Designing and Evaluating Student-facing Learning Dashboards: Lessons Learnt (...
Designing and Evaluating Student-facing Learning Dashboards: Lessons Learnt (...
 
master_thesis.pdf
master_thesis.pdfmaster_thesis.pdf
master_thesis.pdf
 
2018-08-23 EARLI Conference in Bonn Quality Reference Framework for MOOCs Str...
2018-08-23 EARLI Conference in Bonn Quality Reference Framework for MOOCs Str...2018-08-23 EARLI Conference in Bonn Quality Reference Framework for MOOCs Str...
2018-08-23 EARLI Conference in Bonn Quality Reference Framework for MOOCs Str...
 
Interactive recommender systems and dashboards for learning
Interactive recommender systems and dashboards for learningInteractive recommender systems and dashboards for learning
Interactive recommender systems and dashboards for learning
 
2 Studies UX types should know about (Straub UXPA unconference13)
2 Studies UX types should know about (Straub UXPA unconference13)2 Studies UX types should know about (Straub UXPA unconference13)
2 Studies UX types should know about (Straub UXPA unconference13)
 
Data analytics to support awareness and recommendation
Data analytics to support awareness and recommendationData analytics to support awareness and recommendation
Data analytics to support awareness and recommendation
 
Learning to Classify Users in Online Interaction Networks
Learning to Classify Users in Online Interaction NetworksLearning to Classify Users in Online Interaction Networks
Learning to Classify Users in Online Interaction Networks
 

More from Katrien Verbert

Explainability methods
Explainability methodsExplainability methods
Explainability methods
Katrien Verbert
 
Human-centered AI: how can we support end-users to interact with AI?
Human-centered AI: how can we support end-users to interact with AI?Human-centered AI: how can we support end-users to interact with AI?
Human-centered AI: how can we support end-users to interact with AI?
Katrien Verbert
 
Explaining job recommendations: a human-centred perspective
Explaining job recommendations: a human-centred perspectiveExplaining job recommendations: a human-centred perspective
Explaining job recommendations: a human-centred perspective
Katrien Verbert
 
Personalized food recommendations: combining recommendation, visualization an...
Personalized food recommendations: combining recommendation, visualization an...Personalized food recommendations: combining recommendation, visualization an...
Personalized food recommendations: combining recommendation, visualization an...
Katrien Verbert
 
Explaining and Exploring Job Recommendations: a User-driven Approach for Inte...
Explaining and Exploring Job Recommendations: a User-driven Approach for Inte...Explaining and Exploring Job Recommendations: a User-driven Approach for Inte...
Explaining and Exploring Job Recommendations: a User-driven Approach for Inte...
Katrien Verbert
 
Learning analytics for feedback at scale
Learning analytics for feedback at scaleLearning analytics for feedback at scale
Learning analytics for feedback at scale
Katrien Verbert
 
Web Information Systems Lecture 2: HTML
Web Information Systems Lecture 2: HTMLWeb Information Systems Lecture 2: HTML
Web Information Systems Lecture 2: HTML
Katrien Verbert
 
Information Visualisation: perception and principles
Information Visualisation: perception and principlesInformation Visualisation: perception and principles
Information Visualisation: perception and principles
Katrien Verbert
 
Web Information Systems Lecture 1: Introduction
Web Information Systems Lecture 1: IntroductionWeb Information Systems Lecture 1: Introduction
Web Information Systems Lecture 1: Introduction
Katrien Verbert
 
Information Visualisation: Introduction
Information Visualisation: IntroductionInformation Visualisation: Introduction
Information Visualisation: Introduction
Katrien Verbert
 
Student-facing Learning dashboards
Student-facing Learning dashboardsStudent-facing Learning dashboards
Student-facing Learning dashboards
Katrien Verbert
 
EC-TEL 2016 Opening
EC-TEL 2016 OpeningEC-TEL 2016 Opening
EC-TEL 2016 Opening
Katrien Verbert
 
Learning analytics dashboards
Learning analytics dashboardsLearning analytics dashboards
Learning analytics dashboards
Katrien Verbert
 
Open science in the digital humanities
Open science in the digital humanitiesOpen science in the digital humanities
Open science in the digital humanities
Katrien Verbert
 

More from Katrien Verbert (15)

Explainability methods
Explainability methodsExplainability methods
Explainability methods
 
Human-centered AI: how can we support end-users to interact with AI?
Human-centered AI: how can we support end-users to interact with AI?Human-centered AI: how can we support end-users to interact with AI?
Human-centered AI: how can we support end-users to interact with AI?
 
Explaining job recommendations: a human-centred perspective
Explaining job recommendations: a human-centred perspectiveExplaining job recommendations: a human-centred perspective
Explaining job recommendations: a human-centred perspective
 
Personalized food recommendations: combining recommendation, visualization an...
Personalized food recommendations: combining recommendation, visualization an...Personalized food recommendations: combining recommendation, visualization an...
Personalized food recommendations: combining recommendation, visualization an...
 
Explaining and Exploring Job Recommendations: a User-driven Approach for Inte...
Explaining and Exploring Job Recommendations: a User-driven Approach for Inte...Explaining and Exploring Job Recommendations: a User-driven Approach for Inte...
Explaining and Exploring Job Recommendations: a User-driven Approach for Inte...
 
Learning analytics for feedback at scale
Learning analytics for feedback at scaleLearning analytics for feedback at scale
Learning analytics for feedback at scale
 
Web Information Systems Lecture 2: HTML
Web Information Systems Lecture 2: HTMLWeb Information Systems Lecture 2: HTML
Web Information Systems Lecture 2: HTML
 
Information Visualisation: perception and principles
Information Visualisation: perception and principlesInformation Visualisation: perception and principles
Information Visualisation: perception and principles
 
Web Information Systems Lecture 1: Introduction
Web Information Systems Lecture 1: IntroductionWeb Information Systems Lecture 1: Introduction
Web Information Systems Lecture 1: Introduction
 
Information Visualisation: Introduction
Information Visualisation: IntroductionInformation Visualisation: Introduction
Information Visualisation: Introduction
 
Student-facing Learning dashboards
Student-facing Learning dashboardsStudent-facing Learning dashboards
Student-facing Learning dashboards
 
EC-TEL 2016 Opening
EC-TEL 2016 OpeningEC-TEL 2016 Opening
EC-TEL 2016 Opening
 
Learning analytics dashboards
Learning analytics dashboardsLearning analytics dashboards
Learning analytics dashboards
 
Open science in the digital humanities
Open science in the digital humanitiesOpen science in the digital humanities
Open science in the digital humanities
 
Visual analytics
Visual analyticsVisual analytics
Visual analytics
 

Recently uploaded

S.1 chemistry scheme term 2 for ordinary level
S.1 chemistry scheme term 2 for ordinary levelS.1 chemistry scheme term 2 for ordinary level
S.1 chemistry scheme term 2 for ordinary level
ronaldlakony0
 
Earliest Galaxies in the JADES Origins Field: Luminosity Function and Cosmic ...
Earliest Galaxies in the JADES Origins Field: Luminosity Function and Cosmic ...Earliest Galaxies in the JADES Origins Field: Luminosity Function and Cosmic ...
Earliest Galaxies in the JADES Origins Field: Luminosity Function and Cosmic ...
Sérgio Sacani
 
Seminar of U.V. Spectroscopy by SAMIR PANDA
 Seminar of U.V. Spectroscopy by SAMIR PANDA Seminar of U.V. Spectroscopy by SAMIR PANDA
Seminar of U.V. Spectroscopy by SAMIR PANDA
SAMIR PANDA
 
general properties of oerganologametal.ppt
general properties of oerganologametal.pptgeneral properties of oerganologametal.ppt
general properties of oerganologametal.ppt
IqrimaNabilatulhusni
 
GBSN - Microbiology (Lab 4) Culture Media
GBSN - Microbiology (Lab 4) Culture MediaGBSN - Microbiology (Lab 4) Culture Media
GBSN - Microbiology (Lab 4) Culture Media
Areesha Ahmad
 
What is greenhouse gasses and how many gasses are there to affect the Earth.
What is greenhouse gasses and how many gasses are there to affect the Earth.What is greenhouse gasses and how many gasses are there to affect the Earth.
What is greenhouse gasses and how many gasses are there to affect the Earth.
moosaasad1975
 
Toxic effects of heavy metals : Lead and Arsenic
Toxic effects of heavy metals : Lead and ArsenicToxic effects of heavy metals : Lead and Arsenic
Toxic effects of heavy metals : Lead and Arsenic
sanjana502982
 
BLOOD AND BLOOD COMPONENT- introduction to blood physiology
BLOOD AND BLOOD COMPONENT- introduction to blood physiologyBLOOD AND BLOOD COMPONENT- introduction to blood physiology
BLOOD AND BLOOD COMPONENT- introduction to blood physiology
NoelManyise1
 
Mudde & Rovira Kaltwasser. - Populism - a very short introduction [2017].pdf
Mudde & Rovira Kaltwasser. - Populism - a very short introduction [2017].pdfMudde & Rovira Kaltwasser. - Populism - a very short introduction [2017].pdf
Mudde & Rovira Kaltwasser. - Populism - a very short introduction [2017].pdf
frank0071
 
Hemostasis_importance& clinical significance.pptx
Hemostasis_importance& clinical significance.pptxHemostasis_importance& clinical significance.pptx
Hemostasis_importance& clinical significance.pptx
muralinath2
 
原版制作(carleton毕业证书)卡尔顿大学毕业证硕士文凭原版一模一样
原版制作(carleton毕业证书)卡尔顿大学毕业证硕士文凭原版一模一样原版制作(carleton毕业证书)卡尔顿大学毕业证硕士文凭原版一模一样
原版制作(carleton毕业证书)卡尔顿大学毕业证硕士文凭原版一模一样
yqqaatn0
 
in vitro propagation of plants lecture note.pptx
in vitro propagation of plants lecture note.pptxin vitro propagation of plants lecture note.pptx
in vitro propagation of plants lecture note.pptx
yusufzako14
 
Chapter 12 - climate change and the energy crisis
Chapter 12 - climate change and the energy crisisChapter 12 - climate change and the energy crisis
Chapter 12 - climate change and the energy crisis
tonzsalvador2222
 
In silico drugs analogue design: novobiocin analogues.pptx
In silico drugs analogue design: novobiocin analogues.pptxIn silico drugs analogue design: novobiocin analogues.pptx
In silico drugs analogue design: novobiocin analogues.pptx
AlaminAfendy1
 
Unveiling the Energy Potential of Marshmallow Deposits.pdf
Unveiling the Energy Potential of Marshmallow Deposits.pdfUnveiling the Energy Potential of Marshmallow Deposits.pdf
Unveiling the Energy Potential of Marshmallow Deposits.pdf
Erdal Coalmaker
 
Introduction to Mean Field Theory(MFT).pptx
Introduction to Mean Field Theory(MFT).pptxIntroduction to Mean Field Theory(MFT).pptx
Introduction to Mean Field Theory(MFT).pptx
zeex60
 
DERIVATION OF MODIFIED BERNOULLI EQUATION WITH VISCOUS EFFECTS AND TERMINAL V...
DERIVATION OF MODIFIED BERNOULLI EQUATION WITH VISCOUS EFFECTS AND TERMINAL V...DERIVATION OF MODIFIED BERNOULLI EQUATION WITH VISCOUS EFFECTS AND TERMINAL V...
DERIVATION OF MODIFIED BERNOULLI EQUATION WITH VISCOUS EFFECTS AND TERMINAL V...
Wasswaderrick3
 
Observation of Io’s Resurfacing via Plume Deposition Using Ground-based Adapt...
Observation of Io’s Resurfacing via Plume Deposition Using Ground-based Adapt...Observation of Io’s Resurfacing via Plume Deposition Using Ground-based Adapt...
Observation of Io’s Resurfacing via Plume Deposition Using Ground-based Adapt...
Sérgio Sacani
 
Lateral Ventricles.pdf very easy good diagrams comprehensive
Lateral Ventricles.pdf very easy good diagrams comprehensiveLateral Ventricles.pdf very easy good diagrams comprehensive
Lateral Ventricles.pdf very easy good diagrams comprehensive
silvermistyshot
 
Leaf Initiation, Growth and Differentiation.pdf
Leaf Initiation, Growth and Differentiation.pdfLeaf Initiation, Growth and Differentiation.pdf
Leaf Initiation, Growth and Differentiation.pdf
RenuJangid3
 

Recently uploaded (20)

S.1 chemistry scheme term 2 for ordinary level
S.1 chemistry scheme term 2 for ordinary levelS.1 chemistry scheme term 2 for ordinary level
S.1 chemistry scheme term 2 for ordinary level
 
Earliest Galaxies in the JADES Origins Field: Luminosity Function and Cosmic ...
Earliest Galaxies in the JADES Origins Field: Luminosity Function and Cosmic ...Earliest Galaxies in the JADES Origins Field: Luminosity Function and Cosmic ...
Earliest Galaxies in the JADES Origins Field: Luminosity Function and Cosmic ...
 
Seminar of U.V. Spectroscopy by SAMIR PANDA
 Seminar of U.V. Spectroscopy by SAMIR PANDA Seminar of U.V. Spectroscopy by SAMIR PANDA
Seminar of U.V. Spectroscopy by SAMIR PANDA
 
general properties of oerganologametal.ppt
general properties of oerganologametal.pptgeneral properties of oerganologametal.ppt
general properties of oerganologametal.ppt
 
GBSN - Microbiology (Lab 4) Culture Media
GBSN - Microbiology (Lab 4) Culture MediaGBSN - Microbiology (Lab 4) Culture Media
GBSN - Microbiology (Lab 4) Culture Media
 
What is greenhouse gasses and how many gasses are there to affect the Earth.
What is greenhouse gasses and how many gasses are there to affect the Earth.What is greenhouse gasses and how many gasses are there to affect the Earth.
What is greenhouse gasses and how many gasses are there to affect the Earth.
 
Toxic effects of heavy metals : Lead and Arsenic
Toxic effects of heavy metals : Lead and ArsenicToxic effects of heavy metals : Lead and Arsenic
Toxic effects of heavy metals : Lead and Arsenic
 
BLOOD AND BLOOD COMPONENT- introduction to blood physiology
BLOOD AND BLOOD COMPONENT- introduction to blood physiologyBLOOD AND BLOOD COMPONENT- introduction to blood physiology
BLOOD AND BLOOD COMPONENT- introduction to blood physiology
 
Mudde & Rovira Kaltwasser. - Populism - a very short introduction [2017].pdf
Mudde & Rovira Kaltwasser. - Populism - a very short introduction [2017].pdfMudde & Rovira Kaltwasser. - Populism - a very short introduction [2017].pdf
Mudde & Rovira Kaltwasser. - Populism - a very short introduction [2017].pdf
 
Hemostasis_importance& clinical significance.pptx
Hemostasis_importance& clinical significance.pptxHemostasis_importance& clinical significance.pptx
Hemostasis_importance& clinical significance.pptx
 
原版制作(carleton毕业证书)卡尔顿大学毕业证硕士文凭原版一模一样
原版制作(carleton毕业证书)卡尔顿大学毕业证硕士文凭原版一模一样原版制作(carleton毕业证书)卡尔顿大学毕业证硕士文凭原版一模一样
原版制作(carleton毕业证书)卡尔顿大学毕业证硕士文凭原版一模一样
 
in vitro propagation of plants lecture note.pptx
in vitro propagation of plants lecture note.pptxin vitro propagation of plants lecture note.pptx
in vitro propagation of plants lecture note.pptx
 
Chapter 12 - climate change and the energy crisis
Chapter 12 - climate change and the energy crisisChapter 12 - climate change and the energy crisis
Chapter 12 - climate change and the energy crisis
 
In silico drugs analogue design: novobiocin analogues.pptx
In silico drugs analogue design: novobiocin analogues.pptxIn silico drugs analogue design: novobiocin analogues.pptx
In silico drugs analogue design: novobiocin analogues.pptx
 
Unveiling the Energy Potential of Marshmallow Deposits.pdf
Unveiling the Energy Potential of Marshmallow Deposits.pdfUnveiling the Energy Potential of Marshmallow Deposits.pdf
Unveiling the Energy Potential of Marshmallow Deposits.pdf
 
Introduction to Mean Field Theory(MFT).pptx
Introduction to Mean Field Theory(MFT).pptxIntroduction to Mean Field Theory(MFT).pptx
Introduction to Mean Field Theory(MFT).pptx
 
DERIVATION OF MODIFIED BERNOULLI EQUATION WITH VISCOUS EFFECTS AND TERMINAL V...
DERIVATION OF MODIFIED BERNOULLI EQUATION WITH VISCOUS EFFECTS AND TERMINAL V...DERIVATION OF MODIFIED BERNOULLI EQUATION WITH VISCOUS EFFECTS AND TERMINAL V...
DERIVATION OF MODIFIED BERNOULLI EQUATION WITH VISCOUS EFFECTS AND TERMINAL V...
 
Observation of Io’s Resurfacing via Plume Deposition Using Ground-based Adapt...
Observation of Io’s Resurfacing via Plume Deposition Using Ground-based Adapt...Observation of Io’s Resurfacing via Plume Deposition Using Ground-based Adapt...
Observation of Io’s Resurfacing via Plume Deposition Using Ground-based Adapt...
 
Lateral Ventricles.pdf very easy good diagrams comprehensive
Lateral Ventricles.pdf very easy good diagrams comprehensiveLateral Ventricles.pdf very easy good diagrams comprehensive
Lateral Ventricles.pdf very easy good diagrams comprehensive
 
Leaf Initiation, Growth and Differentiation.pdf
Leaf Initiation, Growth and Differentiation.pdfLeaf Initiation, Growth and Differentiation.pdf
Leaf Initiation, Growth and Differentiation.pdf
 

Mixed-initiative recommender systems: towards a next generation of recommender systems through user involvement

  • 1. Mixed-initiative Recommender Systems Katrien Verbert Augment/HCI - KU Leuven @katrien_v
  • 2.
  • 3. Human-Computer Interaction group PhD. researcher Oct. 2003 – Feb. 2008 Post-doc Feb. 2008 – Dec. 2012 Assistant Professor Jan. 2013 – Sept 2014 Assistant Professor Oct. 2014 – Sept. 2018 learning analytics – recommender systems – visualisation Associate Professor Oct. 2018 – …
  • 4. Augment/HCI team Robin De Croon Postdoc researcher Katrien Verbert Associate Professor Francisco Gutiérrez PhD researcher Tom Broos PhD researcher Martijn Millecamp PhD researcher Sven Charleer Postdoc researcher Nyi Nyi Htun Postdoc researcher Houda Lamqaddam PhD researcher Yucheng Jin PhD researcher Oscar Alvarado PhD researcher http://augment.cs.kuleuven.be/ Diego Rojo Carcia PhD researcher
  • 5. Peter Brusliovsky Nava Tintarev Cristina ConatiDenis Parra Collaborations
  • 6.
  • 7. 7
  • 8. Verbert, Katrien; Manouselis, Nikos; Ochoa, Xavier;Wolpers, Martin; Drachsler, Hendrik; Bosnic, Ivana; Duval, Erik. Context- aware recommender systems for learning: a survey and future challenges, IEEE Trans. on Learning Technologies, 18 p. (2012) 8
  • 9. 9
  • 10. Combining recommendation and visualization Core objectives: • make recommendations understandable for users • enable users to steer the recommendation process
  • 11. Flexible interaction with RecSys Research visit ¤ Host: Carnegie Mellon University & University of Pittsburg ¤ Collaboration: John Stamper, Peter Brusilovsky, Denis Parra ¤ Period: April 2012 – June 2012 (3 months) Second post-doctoral fellowship FWO ¤ host university: KU Leuven, Belgium ¤ supervisor: Erik Duval ¤ period: Oct 2012 – Sept 2015 11
  • 12. Overview research topics 12 2012 | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 Learning Analytics - Media Consumption – Research Information Systems - Healthcare
  • 13. Motivation 1 3 ¤ multiple relevance prospects in personalized social tagging systems ¤ community relevance prospects ¤ social relevance prospect ¤ content relevance prospect ¤ existing personalized social systems ¤ do not allow to explore and combine multiple relevance prospects ¤ only one prospect can be explored at a given time
  • 14. Also recommendations è personalized relevance prospect 14
  • 15. Shortcomings recommender systems 15 ¤ cold-start issues ¤ difficult to explain the rationale behind recommendations ¤ user control often limited
  • 16. 16 He, C., Parra, D., & Verbert, K. (2016). Interactive recommender systems: A survey of the state of the art and future research challenges and opportunities. Expert Systems with Applications, 56, 9-27.
  • 17. John O'Donovan, Barry Smyth, Brynjar Gretarsson, Svetlin Bostandjiev, and Tobias Höllerer. 2008. PeerChooser: visual interactive recommendation. CHI '08 Related work: PeerChooser
  • 18. Smallworlds 18 Gretarsson, B., O'Donovan, J., Bostandjiev, S., Hall, C. and Höllerer, T. SmallWorlds: Visualizing Social Recommendations. Comput. Graph. Forum, 29, 3 (2010), 833-842.
  • 20. Contributions ¤ new approach to support exploration, transparency and controllability ¤ recommender systems are shown as agents ¤ in parallel to real users and tags ¤ users can interrelate entities to find items ¤ evaluation study that assesses ¤ effectiveness ¤ probability of item selection 20 Verbert, K., Parra, D., Brusilovsky, P., & Duval, E. (2013). Visualizing recommendations to support exploration, transparency and controllability. In Proceedings of the IUI 2013 international conference on Intelligent user interfaces (pp. 351-362). ACM.
  • 21. Conference Navigator The studies were conducted using Conference Navigator 3 http://halley.exp.sis.pitt.edu/cn3/
  • 22. TalkExplorer - I Entities Tags, Recommender Agents, Users 22
  • 24. Our Assumptions ¤ Items which are relevant in more that one aspect could be more valuable to the users ¤ Displaying multiple aspects of relevance visually is important for the users in the process of item’s exploration 24
  • 25. User study 1 ¤ Setup ¤ supervised user study ¤ 21 participants at UMAP 2012 and ACM Hypertext 2012 conferences ¤ Procedure ¤ Tasks ¤ interact with users and their bookmarks ¤ interact with agents ¤ interact with tags ¤ Post-questionnaire 25
  • 26. Evaluation ¤ Data collection ¤ recordings of voice and screen using camtasia studio ¤ system logs ¤ Measurements ¤ effectiveness: # bookmarked items / #explorations ¤ yield: : # bookmarked items / sum of items in selection set 26
  • 27. Summary results Sign. effect. Sign. yield multiple versus one entity 0.003 <0.001 user versus (user + entity) 0.593 <0.001 agent versus (agent + entity) 0.341 <0.001
  • 28. User study 2 ¤ Setup ¤ Unsupervised user study ¤ Conducted at LAK 2012 and ECTEL 2013 (18 users) ¤ Subjects familiar with visualizations, but not much with RecSys ¤ Procedure ¤ Users were left free to explore the interface. ¤ Interactions were logged ¤ Post-questionnaire 28 Verbert, K., Parra, D., & Brusilovsky, P. (2016). Agents vs. users: Visual recommendation of research talks with multiple dimension of relevance. ACM Transactions on Interactive Intelligent Systems (TIIS), 6(2), 11.
  • 29. Summary results 29 Sign. effect. multiple versus one entity <0.001 user versus (user + entity) 0.3682 agent versus (agent + entity) 0.4426
  • 30. Results of Studies 1 & 2 ¤ Effectiveness increases with intersections of more entities ¤ Effectiveness wasn’t affected in the field study (study 2) ¤ … but exploration distribution was affected 30 Average effectiveness Total number of explorations
  • 31. Drawback: visualizing intersections Venn diagram: more natural way to visualize intersections 31 Clustermap Venn diagram Verbert, K., Parra, D., Brusilovksy, P. (2014). The effect of different set-based visualizations on user exploration of recommendations. In : IntRS@RecSys, 2014 (pp. 37-44).
  • 32. IntersectionExplorer (IEx) 32 Cardoso, B., Sedrakyan, G., Gutiérrez, F., Parra, D., Brusilovsky, P., & Verbert, K. (2018). IntersectionExplorer, a multi-perspective approach for exploring recommendations. International Journal of Human-Computer Studies.
  • 34. Research questions ¤ RQ1: Under which condition may a scalable visualisation increase user acceptance of recommended items? ¤ RQ2: Does a scalable set visualisation increase perceived effectiveness of recommendations. ¤ RQ3: Does a scalable set visualisation increase user trust in recommendations? ¤ RQ4: Does a scalable set visualisation improve user satisfaction with a recommender system? 34
  • 35. Study 1 ¤ Within-subjects study with 20 users ¤ Mean age: 32.9 years; SD: 6.32; female: 3 ¤ Baseline: exploration of recommendations in CN3 ¤ Second condition: exploration of recommendations in IE ¤ Data from two conferences: ¤ EC-TEL 2014 (172 items) ¤ EC-TEL 2015 (112 items) 35
  • 37. Subjective feedback Questionnaire results with statistical significance. Differences between the aspects “Fun” and “Choice satisfaction” were not significant after the Bonferroni-Holm correction. 37
  • 38. Study 2: Digital Humanities 38 ¤ 39 users, less technically oriented ¤ Mean age: 38 years; SD: 10; female: 11 ¤ Data from DH conference: +1000 participants
  • 40. Study 3: IUI 2017 40 ¤ 42 users, technically oriented ¤ Mean age: 32.4 years; SD: 10; female: 17 ¤ Data from IUI conference: 111 accepted papers
  • 43. Study 1 vs Study 2 vs Study 3 ¤ Overall ”augmented agents” were used in all three studies ¤ Participants of study 2 (Digital Humanities) ¤ more interested in content perspective ¤ Rated several dimensions lower (use intention, fun, information sufficiency, control) 43
  • 44. Overview research topics 44 2012 | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 Learning analytics - Media Consumption – Research Information Systems - Healthcare
  • 45. Research questions ¤ How do personal characteristics influence the impact of ¤ RQ1: user controls in terms of diversity and acceptance? ¤ RQ2: visualizations in terms of diversity and acceptance? ¤ RQ3: visualizations + user controls in terms of diversity and acceptance? 45
  • 46. 46
  • 47. Study procedure ¤ Tutorial of study ¤ Pre-study questionnaire: ¤ Demographics ¤ Experience ¤ Visual Memory (VM) Capacity ¤ Musical Sophistication (MS) ¤ Task: Based on a chosen scenario for creating a play-list, explore songs and rate all songs in the final playlist ¤ Post-study questionnaire: ¤ perceived quality ¤ perceived accuracy ¤ perceived diversity ¤ Satisfaction ¤ Effectiveness ¤ choice difficulty
  • 48. Experimental platform The Spotify API allows to generate recommendations ¤ based on up to five favourite artists. ¤ 14 musical attributes in order to describe musical preference 48
  • 49. Dependent variables ¤ Perceived diversity: self-reported measure based on questionnaire ¤ Recommendation acceptance: measured by percentage of liked songs in the play-list 49
  • 50. Personal characteristics ¤ Musical sophistication (MS): ¤ measurement of the ability to engage with music in a flexible, effective and nuanced way (Müllensiefen et al., 2014) ¤ Measured using the Goldsmiths Musical Sophistication Index (Gold-MSI) ¤ Visual memory (VM): ¤ the ability to recall visual patterns (Tintarev and Mastoff, 2016) ¤ Measured by Corsi block-tapping test 50
  • 51. User-centered factors ¤ Exp.1 and Exp.3 used framework of Knijnenburg et al. 2012 ¤ Perceived quality: participants' perceived quality of the recommended songs. ¤ Perceived accuracy: participants' perceived accuracy of the recommended songs according to their preference. ¤ Perceived diversity: the similarity among the recommended songs. ¤ Satisfaction: participants' satisfaction about their chosen recommendations. ¤ Choice difficulty: difficulty of choosing a recommended song. ¤ Effectiveness: usefulness of recommendations generated from systems. ¤ Exp. 2 used the ResQue framework 51
  • 52. Hypotheses ¤ H1: The UI setting (user control, visualization, or both) has a significant effect on recommendation acceptance. ¤ H2: The UI setting (user control, visualization, or both) has a significant effect on perceived diversity. ¤ H3: Visual memory (VM) has a significant effect on recommendation acceptance. ¤ H4: Visual memory (VM) has a significant effect on perceived diversity. ¤ H5: Musical sophistication (MS) has a significant effect on recommendation acceptance. ¤ H6: Musical sophistication (MS) has a significant effect on perceived diversity. 52
  • 54. Experiment 1: Effects of User control Jin, Y., Tintarev, N., & Verbert, K. (2018, September). Effects of personal characteristics on music recommender systems with different levels of controllability. In Proceedings of the 12th ACM Conference on Recommender Systems (pp. 13-21). ACM.
  • 55. User profile (PRO) Algorithm parameters (PAR) Recommendations (REC)
  • 56. Evaluation method ¤ Between-subjects ¤ Independent variable: settings of user control ¤ 2x2x2 factorial design ¤ Dependent variables: ¤ Acceptance (ratings) ¤ Cognitive load (NASA-TLX), MS, VM ¤ Framework Knijnenburg et al.
  • 57. CFA - validity questions Quality Accuracy Effectiveness Choice difficulty Diversity Trust Satisfaction The fit of our SEM model X(2, 98) = 257.410 p <.001 RMSEA= 0.083 CFI = 0.980 TLI = 0.968
  • 58. CFA - validity questions Quality Accuracy Effectiveness Choice difficulty Diversity (low AVE value) Trust (low AVE value) Satisfaction (modification index) The fit of our SEM model X(2, 98) = 257.410 p <.001 RMSEA= 0.083 CFI = 0.980 TLI = 0.968
  • 59. Conclusion ¤ Main effects: from REC to PRO to PAR → higher cognitive load ¤ Two-way interaction: does not necessarily result in higher cognitive load. Adding an additional control component to PAR increases the acceptance. PRO*PAR has less cognitive load than PRO and PAR ¤ Three-way interaction: it increases acceptance, and does not lead to higher cognitive load. Increase interaction times and accuracy ¤ High MS leads to higher quality, and thereby result in higher acceptance 59
  • 60. Experiment 2: effects of visualizations Yucheng Jin, Nava Tintarev, and Katrien Verbert. 2018. Effects of Individual Traits on Diversity-Aware Music Recommender User Interfaces. In Proceedings of the 26th Conference on User Modeling, Adaptation and Personalization (UMAP '18). ACM, New York, NY, USA, 291-299. DOI: https://doi.org/10.1145/3209219.3209225
  • 62. UI Design - Baseline SimBub 62
  • 63. Research Questions 63 ¤ How do personal characteristics influence the impact of ¤ RQ1: user controls in terms of diversity and acceptance? ¤ RQ2: visualizations in terms of diversity and acceptance? ¤ RQ3: visualizations + user controls in terms of diversity and acceptance?
  • 64. Study design ¤ Within-subjects ¤ Independent variable: ¤ Type of bubble charts ¤ Dependent variables: ¤ Perceived diversity ¤ Overall usability ¤ Support to identify blind-spots 64
  • 65. Measurements ¤ VM: Corsi block-tapping test ¤ MS: Goldsmiths Musical Sophistication Index ¤ Support to identify blind spots: #explored genres / #available genres ¤ Perceived diversity: ResQue (item-item diversity, categorial diversity, novelty, and serendipity) ¤ Usability: ResQue (usefulness, user’s attitude, behavioral intention) 65
  • 66. Acceptance 66 H1: The UI setting (visualization) has a significant effect on recommendation acceptance (Cannot accept)
  • 67. Perceived diversity 67 H2: The UI setting (visualization) has a significant effect on perceived diversity (Cannot accept) Mann-Whitney test (U = 2614.00, p = .08)
  • 68. Personal characteristics 68 ComBub supports the participants with high VM to gain higher perceived diversity than SimBub. (H4) ComBub supports the participants with high MS to gain higher perceived diversity than SimBub. (H6)
  • 69. Discussion ¤ In general, visualizing the audio features of music has a limited impact on perceived diversity ¤ Positive correlation between personal characteristics (MS, VM) and the perceived diversity in ComBub 69
  • 70. Experiment 3: Effects of Combining User Control and Visualizations 70
  • 71. User interface ¤ Full Control + ComBub ¤ Full Control + SimBub 71
  • 74. Research Questions 74 ¤ How do personal characteristics influence the impact of ¤ RQ1: user controls in terms of diversity and acceptance? ¤ RQ2: visualizations in terms of diversity and acceptance? ¤ RQ3: visualizations + user controls in terms of diversity and acceptance?
  • 75. Acceptance 75 H1: The UI setting has a significant effect on recommendation acceptance (Cannot accept)
  • 76. Perceived diversity 76 Full Control vs Control+ComBub (M=5.43, SD=1.07, p = .005) Full Control vs Control+SimBub (M=5.22, SD=1.20, p = .035)
  • 77. Personal characteristics 77 Correlation results on acceptance and perceived diversity (* denotes statistical significance at the 0.05 level, i.e., p value< 0.05)
  • 78. Discussion ¤ Adding visualization to Full Control seems to increase diversity significantly. ¤ Control + SimBub ¤ Positive correlation between MS and two dependent variables (acceptance, perceived diversity) ¤ Control + ComBub ¤ Positive correlation between personal characteristics (MS, VM) and the perceived diversity in ComBub ¤ MS also has a positive effect on acceptance ¤ => users with high MS are good at tuning the recommender to find high quality recommendations 78
  • 79. Overview research topics 79 2012 | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 Learning Analytics - Media Consumption – Research Information Systems - Healthcare
  • 80. Additional personal characteristics ¤ Need for cognition, visual literacy, … ¤ In collaboration with Cristina Conati 80
  • 82. Augmented reality 82 Gutiérrez, Francisco, Htun, Nyi Nyi, Charleer, Sven, De Croon, Robin, Verbert, Katrien (2019) Designing Augmented Reality Applications for Personal Health Decision-Making. Proceedings of HICSS-52.
  • 83. Tangible Algorithms ¤ Study with Netflix users ¤ Semiotic inspection ¤ Design workshop ¤ Interviews ¤ Abstract representations ¤ Archetype representations 83 Alvarado, O., Geerts, D. and Verbert, K. Towards Tangible Algorithms: Exploring Algorithmic Experience with Users’ Profiling Representations. Submitted to CHI 2019.