SlideShare ist ein Scribd-Unternehmen logo
1 von 69
Downloaden Sie, um offline zu lesen
Social aspects of interactive RecSys:
Bridging the gap between predictive algorithms
and interactive user interfaces
Denis Parra, Assistant Professor
CS Department
School of Engineering
Pontificia Universidad Católica deChile
SoAPS Workshop at ECIR 2018, March 26th 2018
Funded by :
Personal Introduction
• 2008-2013: PhD at U. of Pittsburgh
• 2013 – now: CS Department, PUC (Santiago, Chile)
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 2
8,280 Km
Outline
• Quick intro to
– RecommenderSystems
– Visualization in RecommenderSystems
• Survey of our work on Visualization and
Interaction on RecSys
– With highlights to social aspects of the results
• Summary & Discussion
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 3
In Collaboration with
• Peter Brusilovsky (University of Pittsburgh, USA)
• Katrien Verbert (KU Leuven, Belgium)
• Christoph Trattner (University of Bergen, Norway)
• Chaoli Wang (Notre Dame University, USA)
• Ivania Donoso (alumni, PUC Chile)
• María Sepúlveda (MSc student, PUC Chile)
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 4
INTRODUCTIONTORECSYS
Recommender Systems
* Danboard (Danbo):Amazon’s cardboard robot, in these slides it represents a
recommender system
*
Recommender Systems (RecSys)
Systems that help (groups of) people to find relevant items in
a crowdeditem or information space(MacNee et al. 2006)
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 6
Why do we care about RecSys?
• Nowadays, several domains& applications require
people to make decisions among a large set of items.
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 7
A lil’ bit of History
• First recommender systems were built at the
beginning of 90’s (Tapestry, GroupLens, Ringo)
• Online contests, such as the Netflix prize, grew the
attention on recommender systems beyond
Computer Science
(2006-2009)
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 8
The Recommendation Problem
• A popular way of presenting the recommendation
problem was rating prediction (Netflix prize)
• How good is my prediction?
Item 1 Item 2 … Item m
User 1 1 5 4
User 2 5 1 ?
…
User n 2 5 ?
Predict!
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 9
Traditional Recommendation Methods
• Without covering all possible methods, the two
most typical classifications on recommender
algorithms are
Classification 1 Classification 2
- Collaborative Filtering
- Content-based Filtering
- Hybrid
- Memory-based
- Model-based
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 10
Collaborative Filtering (User-based KNN)
• Find like-minded people to recommend
5
4
4
2
1
5
4
4
Active
user
User_1
User_2
∑
∑
⊂
⊂
−⋅
+=
)(
)(
),(
)(),(
),(
uneighborsn
uneighborsn nni
u
nuuserSim
rrnuuserSim
riupred2
3
4
2
Item 1
Item 2
Item 3
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 11
Content-Based Filtering
• Can be traced back to techniques from IR, where
the User Profile represents a query.
user_profile = {w_1, w_2, …., w_3} using TF-IDF, weighting
Doc_1 = {w_1, w_2, …., w_3}
Doc_2 = {w_1, w_2, …., w_3}
Doc_3 = {w_1, w_2, …., w_3}
Doc_n = {w_1, w_2, …., w_3}
5
4
5
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 12
Hybridization
• Combine previous methods to overcome their
weaknesses (Burke, 2002)
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 13
Model-based: Matrix Factorization
Latent vector of the item
Latent vector of the user
SVD ~
Singular Value
Decomposition
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 14
Other paradigms and techniques
• Recommendation as a graph problem:
– PersonalizedPageRank (Kamvar et al, 2010), (Santos et
al 2016), etc.
• Recommendation as a ranking problem:
– Karatzoglou et al. (2013), Shi et al. (2014), Macedo et al.
(2015), etc.
• Deep learning methods:
– MetaProd2Vecby Vasile et al. (2016), YouTube
Recommendationsby Covington et al. (2016), etc.
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 15
Are accuracy metrics the goal of RecSys?
• Some works started pointing out that small
improvements in RMSE did not have a
proportional improvement in user satisfaction
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 16
MovieLens:
Traditional RecSys Interface
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 17
PeerChooser (2008) Controllability in CF
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 18
O’Donovan et al. “PeerChooser: Visual Interactive Recommendation” (2008)
SmallWorlds: Expanded Explainability
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 19
Gretarsson et al. “SmallWorlds:Visualizing social recommendations” (2010)
TasteWeights: Hybrid Control and Inspect
Bostandjev et al. “TasteWeights:A Visual Interactive Hybrid Recommender System” (2012)
Controllability:
Sliders that let users
control the
importance of
preferences and
contexts
Inspectability: lines
that connect
recommended items
with contexts and user
preferences
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 20
IUI 2017
• Loepp et al. (2017)
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 21
More Details? Check our survey
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 22
He, C., Parra, D., & Verbert, K. (2016). Interactive recommender
systems: a survey of the state of the art and future research
challenges andopportunities.Expert Systems with Applications, 56, 9-27.
My Take on RecSys Research (2009 ~)
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 23
IUI 2013 IUI 2014 UMAP 2016
E.I.-VDA 2018 IUI 2018
Visual & interactive RecSys interfaces
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 24
CNVis EpistAid
TalkExplorer SetFusion Moodplay
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 25
TalkExplorer – IUI 2013
• Adaptation of Aduna Visualization to CN
• Main researchquestion: Does fusion (intersection) of
contextsof relevance improve user experience with RecSys?
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 26
Research Platform
• The studies were conducted using Conference
Navigator, a Conference Support System
• Our goal was recommending conference talks
Program Proceedings Author List Recommendations
http://halley.exp.sis.pitt.edu/cn3/
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 27
TalkExplorer - I
Entities
Tags, Recommender Agents,
Users
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 28
TalkExplorer - II
Recommender
Recommender
Cluster
with
intersecti
on of
entities Cluster (of
talks)
associated to
only one
entity
• Canvas Area: Intersections of Different Entities
User
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 29
TalkExplorer - III
Items
Talks explored by the
user
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 30
Evaluation: Intersections & Effectiveness
• What do we call an “Intersection”?
• We used #explorations on intersections and their
effectiveness, defined as:
Effectiveness =
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 31
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
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 32
TalkExplorer Studies I & II
• Study I
– ControlledExperiment:Userswere asked todiscover
relevant talksby exploringthe three types of entities:tags,
recommenderagents and users.
– Conductedat HypertextandUMAP 2012(21 users)
– Subjectsfamiliarwith Visualizationsand Recsys
• Study II
– FieldStudy: Users were left free to explore the interface.
– Conductedat LAK 2012 and ECTEL 2013 (18 users)
– Subjectsfamiliarwith visualizations, but not much with
RecSys
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 33
Results of Studies I & II
• Effectiveness increases
with intersections of
more entities
• Effectiveness wasn’t
affected in the field
study (study 2)
• … but exploration
distribution was
affected
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 34
More detail on entities
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 35
More detail on entities
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 36
• Despite not being the most effective, the
user (social) entity attracted by far more
explorations from users.
Social ~ Trust in RecSys
• O'Donovan, John, and Barry Smyth. "Trust in
recommender systems," IUI, 2005.
• Golbeck, Jennifer, and James Hendler. "Filmtrust:
Movie recommendations using trust in web-based
social networks," ComNet, 2006.
• Jamali, Mohsen, and Martin Ester. "A matrix
factorization technique with trust propagation for
recommendation in social networks." RecSys, 2010.
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 37
SetFusion – IUI 2014
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 38
SetFusion I
Traditional
Ranked List
Papers sorted by
Relevance.
It combines 3
recommendation
approaches.
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 39
SetFusion - II
Sliders
Allow the user to control the importance of
each data source or recommendation method
Interactive Venn Diagram
Allows the user to inspect and to filter papers
recommended. Actions available:
- Filter item list by clicking on an area
- Highlight a paper by mouse-over on a circle
- Scroll to paper by clicking on a circle
- Indicate bookmarked papers
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 40
Study : iConference
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 41
A
B
C
• A and C: Social
• B: Content-based
Rating per method – Effect of Visuals
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 42
Rating per method – Effect of Visuals
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 43
No significant difference
Rating per method – Effect of Visuals
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 44
Hybrid algorithm + visualization
yield the only significant
difference
Summary & Conclusions
• The combination of several sources of relevance
has an impact on recommendation, being the social
aspect among the most relevant.
• The visual paradigm combined with social aspects
used can have a significant effect on user behavior.
• We need to keep working on visual
representations that promote exploration without
decreasing recommendation accuracy.
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 45
MOODFROM SOCIAL SOURCES FOR
MUSICRECOMMENDATION
Moodplay
• MoodPlay
– With Ivana Andjelkovic& John O’Donovan (UCSB)
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 47
Andjelkovic,I., Parra, D., & O'Donovan,J. (2016). Moodplay:Interactive
Mood-based Music Discovery and Recommendation.In Proceedings of
the 2016 Conference on User Modeling Adaptation and Personalization (pp.
275-279).ACM.
MoodPlay
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 48
System Architecture
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 49
Emotion Models
• Modelo de emociones de Russel
(1980)
• GEMS (2008)
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 50
Moods and Music: the GEMS model
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 51
Components
• Colors: GEMS emotions
• Dots: artists
• Zooming and Panning
• Playing &
Recommendations
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 52
Components
• Colors: GEMS emotions
• Dots: artists
• Zooming and Panning
• Playing &
Recommendations
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 53
Components
• Colors: GEMS emotions
• Dots: artists
• Zooming and Panning
• Playing &
Recommendations
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 54
Components
• Colors: GEMS emotions
• Dots: artists
• Zooming and Panning
• Playing &
Recommendations
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 55
Emotion-Aware Recommendation
• Using several Web APIs, we
collected users’ perception of mood
associated to artists.
• Then, using artists as input, we
calculate recommendations of
similarly perceived artists
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 56
MoodPlay
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 57
• Our results indicate that users’ reported mood and
artist mood have an effect on people satisfaction
with the system.
• Try it at
http://moodplay.pythonanywhere.com
SOME CHALLENGES &
CONCLUSION
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 59
Interactive Relevance Feedback
Interface for evidence–based
Medicine
Ivania Donoso-Guzmán
Denis Parra
Honorable mention for best paper award
ACM IUI 2018
Belleret al (2013).
… takes from 6 to 12 meses
Process for answering a clinical
question
60
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018)
61
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018)
EpistAid
Model Interfaz N
Documentos Vistos Ground Truth
Recall Precision F-1 score Recall
BM25 Non-Viz 12 0.66+/-0.08 0.52 +/- 0.06 0.58 +/- 0.07 0.20 +/- 0.04
BM25 Viz 11 0.71+/-0.06 0.64 +/- 0.04 0.64 +/- 0.03 0.18 +/- 0.02
Rocchio Non-Viz 11 0.65+/- 0.08 0.73 +/- 0.02 0.65 +/- 0.05 0.21 +/- 0.04
Rocchio Viz 12 0.77 +/- 0.06 0.67 +/- 0.01 0.70 +/- 0.03 0.23 +/- 0.03
Promising Results
62
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018)
Future Work
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 63
- CollaborativeModel:
- Whichis the best way to providea collaborative
human-in-the-loopinterfacefor Evidence-Based
Medicine?
- … Ideas from research on collaborative web
search, for instance:
Yue et al. “Influences on query reformulation in
collaborative web search”. IEEE Computer Magazine,
2014.
IEEE VIS 2017 – Panel ML & Vis ?
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 64
Deep Learning and Social on RecSys
• Lei, Chenyi, et al. "Comparative deep learning of
hybrid representations for image
recommendations." CVPR. 2016.
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 65
Deep Learning and Social on RecSys
• Lei, Chenyi, et al. "Comparative deep learning of
hybrid representations for image
recommendations." CVPR. 2016.
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 66
- How to make better
sense of the social
embeddings for
explaining
recommendations?
IEEE VIS 2017 – Panel ML & Vis ?
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 67
Conclusion
March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 68
• The social aspects are indeed relevant in
recommendation systems, and their effect on
prediction accuracyhas been already studied.
• The studies presented in this talk show that combining
interactive visualizations with social and other
relevance signal can have an important effect on users’
perception of recommendations.
• I invite researcher to further study the connections
between social aspects, visualization and their effect
on recommender systems.
THANKS!
dparra@ing.puc.cl

Weitere ähnliche Inhalte

Ähnlich wie Social Aspects of Interactive Recommender Systems

Rubrics for DMPs
Rubrics for DMPsRubrics for DMPs
Rubrics for DMPsJisc RDM
 
Prototyping: 4 Strategic Factors for Design Teams
Prototyping: 4 Strategic Factors for Design TeamsPrototyping: 4 Strategic Factors for Design Teams
Prototyping: 4 Strategic Factors for Design TeamsLyle Kantrovich
 
User Studies for APG: How to support system development with user feedback?
User Studies for APG: How to support system development with user feedback?User Studies for APG: How to support system development with user feedback?
User Studies for APG: How to support system development with user feedback?Joni Salminen
 
Tutorial on Bias in Rec Sys @ UMAP2020
Tutorial on Bias in Rec Sys @ UMAP2020Tutorial on Bias in Rec Sys @ UMAP2020
Tutorial on Bias in Rec Sys @ UMAP2020Mirko Marras
 
180719 테헤란로 런치클럽-실무자 관점에서 본 Spotify의 유저리서치 방식 (Spotify 백원희님)
180719 테헤란로 런치클럽-실무자 관점에서 본 Spotify의 유저리서치 방식 (Spotify 백원희님)180719 테헤란로 런치클럽-실무자 관점에서 본 Spotify의 유저리서치 방식 (Spotify 백원희님)
180719 테헤란로 런치클럽-실무자 관점에서 본 Spotify의 유저리서치 방식 (Spotify 백원희님)StartupAlliance
 
Wimmics Research Team 2015 Activity Report
Wimmics Research Team 2015 Activity ReportWimmics Research Team 2015 Activity Report
Wimmics Research Team 2015 Activity ReportFabien Gandon
 
Evalita2018 iListen - itaLIan Speech acT labEliNg
Evalita2018 iListen - itaLIan Speech acT labEliNgEvalita2018 iListen - itaLIan Speech acT labEliNg
Evalita2018 iListen - itaLIan Speech acT labEliNgNicole Novielli
 
Social media data stewardship: The ethics of social media data use for research
Social media data stewardship: The ethics of social media data use for researchSocial media data stewardship: The ethics of social media data use for research
Social media data stewardship: The ethics of social media data use for researchToronto Metropolitan University
 
OpenAIRE: Implementing Open Science
OpenAIRE: Implementing Open ScienceOpenAIRE: Implementing Open Science
OpenAIRE: Implementing Open Sciencee-ROSA
 
Approaches of Data Analysis: Networks generated through Social Media
Approaches of Data Analysis: Networks generated through Social MediaApproaches of Data Analysis: Networks generated through Social Media
Approaches of Data Analysis: Networks generated through Social MediaJanna Joceli Omena
 
Foresight by Online Communities - The Case of Renewable Energies
Foresight by Online Communities - The Case of Renewable EnergiesForesight by Online Communities - The Case of Renewable Energies
Foresight by Online Communities - The Case of Renewable EnergiesMichael Andreas Zeng
 
Metadata mapping and vocabulary: consistency for all in scholarly communicati...
Metadata mapping and vocabulary: consistency for all in scholarly communicati...Metadata mapping and vocabulary: consistency for all in scholarly communicati...
Metadata mapping and vocabulary: consistency for all in scholarly communicati...CILIP MDG
 
Data Science Skills Study 2018 by AIM & Great Learning
Data Science Skills Study 2018 by AIM & Great LearningData Science Skills Study 2018 by AIM & Great Learning
Data Science Skills Study 2018 by AIM & Great LearningAnalytics India Magazine
 
New Data for Innovation Policy
New Data for Innovation PolicyNew Data for Innovation Policy
New Data for Innovation PolicyJuan Mateos-Garcia
 
Garcia - New data for innovation policy
Garcia - New data for innovation policyGarcia - New data for innovation policy
Garcia - New data for innovation policyinnovationoecd
 
Measuring Impact: Towards a data citation metric
Measuring Impact: Towards a data citation metricMeasuring Impact: Towards a data citation metric
Measuring Impact: Towards a data citation metricEdward Baker
 
Metadata 2020: Projects
Metadata 2020: ProjectsMetadata 2020: Projects
Metadata 2020: ProjectsNASIG
 
GIJC19 - NodeXL Tutorial - Session 1
GIJC19 - NodeXL Tutorial - Session 1GIJC19 - NodeXL Tutorial - Session 1
GIJC19 - NodeXL Tutorial - Session 1Harald Meier
 

Ähnlich wie Social Aspects of Interactive Recommender Systems (20)

Rubrics for DMPs
Rubrics for DMPsRubrics for DMPs
Rubrics for DMPs
 
Prototyping: 4 Strategic Factors for Design Teams
Prototyping: 4 Strategic Factors for Design TeamsPrototyping: 4 Strategic Factors for Design Teams
Prototyping: 4 Strategic Factors for Design Teams
 
User Studies for APG: How to support system development with user feedback?
User Studies for APG: How to support system development with user feedback?User Studies for APG: How to support system development with user feedback?
User Studies for APG: How to support system development with user feedback?
 
Tutorial on Bias in Rec Sys @ UMAP2020
Tutorial on Bias in Rec Sys @ UMAP2020Tutorial on Bias in Rec Sys @ UMAP2020
Tutorial on Bias in Rec Sys @ UMAP2020
 
180719 테헤란로 런치클럽-실무자 관점에서 본 Spotify의 유저리서치 방식 (Spotify 백원희님)
180719 테헤란로 런치클럽-실무자 관점에서 본 Spotify의 유저리서치 방식 (Spotify 백원희님)180719 테헤란로 런치클럽-실무자 관점에서 본 Spotify의 유저리서치 방식 (Spotify 백원희님)
180719 테헤란로 런치클럽-실무자 관점에서 본 Spotify의 유저리서치 방식 (Spotify 백원희님)
 
Let’s All Agree on What We’re Counting and How
Let’s All Agree on What We’re Counting and HowLet’s All Agree on What We’re Counting and How
Let’s All Agree on What We’re Counting and How
 
Wimmics Research Team 2015 Activity Report
Wimmics Research Team 2015 Activity ReportWimmics Research Team 2015 Activity Report
Wimmics Research Team 2015 Activity Report
 
Evalita2018 iListen - itaLIan Speech acT labEliNg
Evalita2018 iListen - itaLIan Speech acT labEliNgEvalita2018 iListen - itaLIan Speech acT labEliNg
Evalita2018 iListen - itaLIan Speech acT labEliNg
 
Social media data stewardship: The ethics of social media data use for research
Social media data stewardship: The ethics of social media data use for researchSocial media data stewardship: The ethics of social media data use for research
Social media data stewardship: The ethics of social media data use for research
 
OpenAIRE: Implementing Open Science
OpenAIRE: Implementing Open ScienceOpenAIRE: Implementing Open Science
OpenAIRE: Implementing Open Science
 
Approaches of Data Analysis: Networks generated through Social Media
Approaches of Data Analysis: Networks generated through Social MediaApproaches of Data Analysis: Networks generated through Social Media
Approaches of Data Analysis: Networks generated through Social Media
 
Toward supporting decision-making under uncertainty in digital humanities wit...
Toward supporting decision-making under uncertainty in digital humanities wit...Toward supporting decision-making under uncertainty in digital humanities wit...
Toward supporting decision-making under uncertainty in digital humanities wit...
 
Foresight by Online Communities - The Case of Renewable Energies
Foresight by Online Communities - The Case of Renewable EnergiesForesight by Online Communities - The Case of Renewable Energies
Foresight by Online Communities - The Case of Renewable Energies
 
Metadata mapping and vocabulary: consistency for all in scholarly communicati...
Metadata mapping and vocabulary: consistency for all in scholarly communicati...Metadata mapping and vocabulary: consistency for all in scholarly communicati...
Metadata mapping and vocabulary: consistency for all in scholarly communicati...
 
Data Science Skills Study 2018 by AIM & Great Learning
Data Science Skills Study 2018 by AIM & Great LearningData Science Skills Study 2018 by AIM & Great Learning
Data Science Skills Study 2018 by AIM & Great Learning
 
New Data for Innovation Policy
New Data for Innovation PolicyNew Data for Innovation Policy
New Data for Innovation Policy
 
Garcia - New data for innovation policy
Garcia - New data for innovation policyGarcia - New data for innovation policy
Garcia - New data for innovation policy
 
Measuring Impact: Towards a data citation metric
Measuring Impact: Towards a data citation metricMeasuring Impact: Towards a data citation metric
Measuring Impact: Towards a data citation metric
 
Metadata 2020: Projects
Metadata 2020: ProjectsMetadata 2020: Projects
Metadata 2020: Projects
 
GIJC19 - NodeXL Tutorial - Session 1
GIJC19 - NodeXL Tutorial - Session 1GIJC19 - NodeXL Tutorial - Session 1
GIJC19 - NodeXL Tutorial - Session 1
 

Mehr von Denis Parra Santander

Human-Centered Machine Learning: Harnessing Visualization and Interactivity f...
Human-Centered Machine Learning: Harnessing Visualization and Interactivity f...Human-Centered Machine Learning: Harnessing Visualization and Interactivity f...
Human-Centered Machine Learning: Harnessing Visualization and Interactivity f...Denis Parra Santander
 
Data Fusion for Dealing with the Recommendation Problem
Data Fusion for Dealing with the Recommendation ProblemData Fusion for Dealing with the Recommendation Problem
Data Fusion for Dealing with the Recommendation ProblemDenis Parra Santander
 
Identifying Relevant Messages in a Twitter-based Citizen Channel for Natural ...
Identifying Relevant Messages in a Twitter-based Citizen Channel for Natural ...Identifying Relevant Messages in a Twitter-based Citizen Channel for Natural ...
Identifying Relevant Messages in a Twitter-based Citizen Channel for Natural ...Denis Parra Santander
 
Research on Recommender Systems: Beyond Ratings and Lists
Research on Recommender Systems: Beyond Ratings and ListsResearch on Recommender Systems: Beyond Ratings and Lists
Research on Recommender Systems: Beyond Ratings and ListsDenis Parra Santander
 
The Effect of Different Set-based Visualizations on User Exploration of Reco...
The Effect of Different Set-based  Visualizations on User Exploration of Reco...The Effect of Different Set-based  Visualizations on User Exploration of Reco...
The Effect of Different Set-based Visualizations on User Exploration of Reco...Denis Parra Santander
 
SetFusion Visual Hybrid Recommender - IUI 2014
SetFusion Visual Hybrid Recommender -  IUI 2014SetFusion Visual Hybrid Recommender -  IUI 2014
SetFusion Visual Hybrid Recommender - IUI 2014Denis Parra Santander
 
Walk the Talk: Analyzing the relation between implicit and explicit feedback ...
Walk the Talk: Analyzing the relation between implicit and explicit feedback ...Walk the Talk: Analyzing the relation between implicit and explicit feedback ...
Walk the Talk: Analyzing the relation between implicit and explicit feedback ...Denis Parra Santander
 
Network Visualization guest lecture at #DataVizQMSS at @Columbia / #SNA at PU...
Network Visualization guest lecture at #DataVizQMSS at @Columbia / #SNA at PU...Network Visualization guest lecture at #DataVizQMSS at @Columbia / #SNA at PU...
Network Visualization guest lecture at #DataVizQMSS at @Columbia / #SNA at PU...Denis Parra Santander
 
A Hybrid Peer Recommender System for a Online Community Teachers
A Hybrid Peer Recommender System for a Online Community TeachersA Hybrid Peer Recommender System for a Online Community Teachers
A Hybrid Peer Recommender System for a Online Community TeachersDenis Parra Santander
 
Implicit Feedback Recommendation via Implicit-to-Explicit Ordinal Logistic Re...
Implicit Feedback Recommendation via Implicit-to-Explicit Ordinal Logistic Re...Implicit Feedback Recommendation via Implicit-to-Explicit Ordinal Logistic Re...
Implicit Feedback Recommendation via Implicit-to-Explicit Ordinal Logistic Re...Denis Parra Santander
 
Currents steps to be a researcher and faculty
Currents steps to be a researcher and facultyCurrents steps to be a researcher and faculty
Currents steps to be a researcher and facultyDenis Parra Santander
 
Evaluation of Collaborative Filtering Algorithms for Recommending Articles on...
Evaluation of Collaborative Filtering Algorithms for Recommending Articles on...Evaluation of Collaborative Filtering Algorithms for Recommending Articles on...
Evaluation of Collaborative Filtering Algorithms for Recommending Articles on...Denis Parra Santander
 

Mehr von Denis Parra Santander (14)

Human-Centered Machine Learning: Harnessing Visualization and Interactivity f...
Human-Centered Machine Learning: Harnessing Visualization and Interactivity f...Human-Centered Machine Learning: Harnessing Visualization and Interactivity f...
Human-Centered Machine Learning: Harnessing Visualization and Interactivity f...
 
Data Fusion for Dealing with the Recommendation Problem
Data Fusion for Dealing with the Recommendation ProblemData Fusion for Dealing with the Recommendation Problem
Data Fusion for Dealing with the Recommendation Problem
 
LDA on social bookmarking systems
LDA on social bookmarking systemsLDA on social bookmarking systems
LDA on social bookmarking systems
 
Identifying Relevant Messages in a Twitter-based Citizen Channel for Natural ...
Identifying Relevant Messages in a Twitter-based Citizen Channel for Natural ...Identifying Relevant Messages in a Twitter-based Citizen Channel for Natural ...
Identifying Relevant Messages in a Twitter-based Citizen Channel for Natural ...
 
Research on Recommender Systems: Beyond Ratings and Lists
Research on Recommender Systems: Beyond Ratings and ListsResearch on Recommender Systems: Beyond Ratings and Lists
Research on Recommender Systems: Beyond Ratings and Lists
 
The Effect of Different Set-based Visualizations on User Exploration of Reco...
The Effect of Different Set-based  Visualizations on User Exploration of Reco...The Effect of Different Set-based  Visualizations on User Exploration of Reco...
The Effect of Different Set-based Visualizations on User Exploration of Reco...
 
Twitter in Academic Conferences
Twitter in Academic ConferencesTwitter in Academic Conferences
Twitter in Academic Conferences
 
SetFusion Visual Hybrid Recommender - IUI 2014
SetFusion Visual Hybrid Recommender -  IUI 2014SetFusion Visual Hybrid Recommender -  IUI 2014
SetFusion Visual Hybrid Recommender - IUI 2014
 
Walk the Talk: Analyzing the relation between implicit and explicit feedback ...
Walk the Talk: Analyzing the relation between implicit and explicit feedback ...Walk the Talk: Analyzing the relation between implicit and explicit feedback ...
Walk the Talk: Analyzing the relation between implicit and explicit feedback ...
 
Network Visualization guest lecture at #DataVizQMSS at @Columbia / #SNA at PU...
Network Visualization guest lecture at #DataVizQMSS at @Columbia / #SNA at PU...Network Visualization guest lecture at #DataVizQMSS at @Columbia / #SNA at PU...
Network Visualization guest lecture at #DataVizQMSS at @Columbia / #SNA at PU...
 
A Hybrid Peer Recommender System for a Online Community Teachers
A Hybrid Peer Recommender System for a Online Community TeachersA Hybrid Peer Recommender System for a Online Community Teachers
A Hybrid Peer Recommender System for a Online Community Teachers
 
Implicit Feedback Recommendation via Implicit-to-Explicit Ordinal Logistic Re...
Implicit Feedback Recommendation via Implicit-to-Explicit Ordinal Logistic Re...Implicit Feedback Recommendation via Implicit-to-Explicit Ordinal Logistic Re...
Implicit Feedback Recommendation via Implicit-to-Explicit Ordinal Logistic Re...
 
Currents steps to be a researcher and faculty
Currents steps to be a researcher and facultyCurrents steps to be a researcher and faculty
Currents steps to be a researcher and faculty
 
Evaluation of Collaborative Filtering Algorithms for Recommending Articles on...
Evaluation of Collaborative Filtering Algorithms for Recommending Articles on...Evaluation of Collaborative Filtering Algorithms for Recommending Articles on...
Evaluation of Collaborative Filtering Algorithms for Recommending Articles on...
 

Kürzlich hochgeladen

Salesforce Miami User Group Event - 1st Quarter 2024
Salesforce Miami User Group Event - 1st Quarter 2024Salesforce Miami User Group Event - 1st Quarter 2024
Salesforce Miami User Group Event - 1st Quarter 2024SkyPlanner
 
Building Your Own AI Instance (TBLC AI )
Building Your Own AI Instance (TBLC AI )Building Your Own AI Instance (TBLC AI )
Building Your Own AI Instance (TBLC AI )Brian Pichman
 
Empowering Africa's Next Generation: The AI Leadership Blueprint
Empowering Africa's Next Generation: The AI Leadership BlueprintEmpowering Africa's Next Generation: The AI Leadership Blueprint
Empowering Africa's Next Generation: The AI Leadership BlueprintMahmoud Rabie
 
100+ ChatGPT Prompts for SEO Optimization
100+ ChatGPT Prompts for SEO Optimization100+ ChatGPT Prompts for SEO Optimization
100+ ChatGPT Prompts for SEO Optimizationarrow10202532yuvraj
 
UiPath Studio Web workshop series - Day 5
UiPath Studio Web workshop series - Day 5UiPath Studio Web workshop series - Day 5
UiPath Studio Web workshop series - Day 5DianaGray10
 
UiPath Platform: The Backend Engine Powering Your Automation - Session 1
UiPath Platform: The Backend Engine Powering Your Automation - Session 1UiPath Platform: The Backend Engine Powering Your Automation - Session 1
UiPath Platform: The Backend Engine Powering Your Automation - Session 1DianaGray10
 
Nanopower In Semiconductor Industry.pdf
Nanopower  In Semiconductor Industry.pdfNanopower  In Semiconductor Industry.pdf
Nanopower In Semiconductor Industry.pdfPedro Manuel
 
Secure your environment with UiPath and CyberArk technologies - Session 1
Secure your environment with UiPath and CyberArk technologies - Session 1Secure your environment with UiPath and CyberArk technologies - Session 1
Secure your environment with UiPath and CyberArk technologies - Session 1DianaGray10
 
UiPath Clipboard AI: "A TIME Magazine Best Invention of 2023 Unveiled"
UiPath Clipboard AI: "A TIME Magazine Best Invention of 2023 Unveiled"UiPath Clipboard AI: "A TIME Magazine Best Invention of 2023 Unveiled"
UiPath Clipboard AI: "A TIME Magazine Best Invention of 2023 Unveiled"DianaGray10
 
20230202 - Introduction to tis-py
20230202 - Introduction to tis-py20230202 - Introduction to tis-py
20230202 - Introduction to tis-pyJamie (Taka) Wang
 
UiPath Studio Web workshop series - Day 8
UiPath Studio Web workshop series - Day 8UiPath Studio Web workshop series - Day 8
UiPath Studio Web workshop series - Day 8DianaGray10
 
AI Fame Rush Review – Virtual Influencer Creation In Just Minutes
AI Fame Rush Review – Virtual Influencer Creation In Just MinutesAI Fame Rush Review – Virtual Influencer Creation In Just Minutes
AI Fame Rush Review – Virtual Influencer Creation In Just MinutesMd Hossain Ali
 
The Kubernetes Gateway API and its role in Cloud Native API Management
The Kubernetes Gateway API and its role in Cloud Native API ManagementThe Kubernetes Gateway API and its role in Cloud Native API Management
The Kubernetes Gateway API and its role in Cloud Native API ManagementNuwan Dias
 
Computer 10: Lesson 10 - Online Crimes and Hazards
Computer 10: Lesson 10 - Online Crimes and HazardsComputer 10: Lesson 10 - Online Crimes and Hazards
Computer 10: Lesson 10 - Online Crimes and HazardsSeth Reyes
 
COMPUTER 10: Lesson 7 - File Storage and Online Collaboration
COMPUTER 10: Lesson 7 - File Storage and Online CollaborationCOMPUTER 10: Lesson 7 - File Storage and Online Collaboration
COMPUTER 10: Lesson 7 - File Storage and Online Collaborationbruanjhuli
 
Using IESVE for Loads, Sizing and Heat Pump Modeling to Achieve Decarbonization
Using IESVE for Loads, Sizing and Heat Pump Modeling to Achieve DecarbonizationUsing IESVE for Loads, Sizing and Heat Pump Modeling to Achieve Decarbonization
Using IESVE for Loads, Sizing and Heat Pump Modeling to Achieve DecarbonizationIES VE
 
Anypoint Code Builder , Google Pub sub connector and MuleSoft RPA
Anypoint Code Builder , Google Pub sub connector and MuleSoft RPAAnypoint Code Builder , Google Pub sub connector and MuleSoft RPA
Anypoint Code Builder , Google Pub sub connector and MuleSoft RPAshyamraj55
 
IEEE Computer Society’s Strategic Activities and Products including SWEBOK Guide
IEEE Computer Society’s Strategic Activities and Products including SWEBOK GuideIEEE Computer Society’s Strategic Activities and Products including SWEBOK Guide
IEEE Computer Society’s Strategic Activities and Products including SWEBOK GuideHironori Washizaki
 
UiPath Community: AI for UiPath Automation Developers
UiPath Community: AI for UiPath Automation DevelopersUiPath Community: AI for UiPath Automation Developers
UiPath Community: AI for UiPath Automation DevelopersUiPathCommunity
 

Kürzlich hochgeladen (20)

Salesforce Miami User Group Event - 1st Quarter 2024
Salesforce Miami User Group Event - 1st Quarter 2024Salesforce Miami User Group Event - 1st Quarter 2024
Salesforce Miami User Group Event - 1st Quarter 2024
 
Building Your Own AI Instance (TBLC AI )
Building Your Own AI Instance (TBLC AI )Building Your Own AI Instance (TBLC AI )
Building Your Own AI Instance (TBLC AI )
 
Empowering Africa's Next Generation: The AI Leadership Blueprint
Empowering Africa's Next Generation: The AI Leadership BlueprintEmpowering Africa's Next Generation: The AI Leadership Blueprint
Empowering Africa's Next Generation: The AI Leadership Blueprint
 
100+ ChatGPT Prompts for SEO Optimization
100+ ChatGPT Prompts for SEO Optimization100+ ChatGPT Prompts for SEO Optimization
100+ ChatGPT Prompts for SEO Optimization
 
UiPath Studio Web workshop series - Day 5
UiPath Studio Web workshop series - Day 5UiPath Studio Web workshop series - Day 5
UiPath Studio Web workshop series - Day 5
 
UiPath Platform: The Backend Engine Powering Your Automation - Session 1
UiPath Platform: The Backend Engine Powering Your Automation - Session 1UiPath Platform: The Backend Engine Powering Your Automation - Session 1
UiPath Platform: The Backend Engine Powering Your Automation - Session 1
 
Nanopower In Semiconductor Industry.pdf
Nanopower  In Semiconductor Industry.pdfNanopower  In Semiconductor Industry.pdf
Nanopower In Semiconductor Industry.pdf
 
Secure your environment with UiPath and CyberArk technologies - Session 1
Secure your environment with UiPath and CyberArk technologies - Session 1Secure your environment with UiPath and CyberArk technologies - Session 1
Secure your environment with UiPath and CyberArk technologies - Session 1
 
UiPath Clipboard AI: "A TIME Magazine Best Invention of 2023 Unveiled"
UiPath Clipboard AI: "A TIME Magazine Best Invention of 2023 Unveiled"UiPath Clipboard AI: "A TIME Magazine Best Invention of 2023 Unveiled"
UiPath Clipboard AI: "A TIME Magazine Best Invention of 2023 Unveiled"
 
20230202 - Introduction to tis-py
20230202 - Introduction to tis-py20230202 - Introduction to tis-py
20230202 - Introduction to tis-py
 
UiPath Studio Web workshop series - Day 8
UiPath Studio Web workshop series - Day 8UiPath Studio Web workshop series - Day 8
UiPath Studio Web workshop series - Day 8
 
AI Fame Rush Review – Virtual Influencer Creation In Just Minutes
AI Fame Rush Review – Virtual Influencer Creation In Just MinutesAI Fame Rush Review – Virtual Influencer Creation In Just Minutes
AI Fame Rush Review – Virtual Influencer Creation In Just Minutes
 
The Kubernetes Gateway API and its role in Cloud Native API Management
The Kubernetes Gateway API and its role in Cloud Native API ManagementThe Kubernetes Gateway API and its role in Cloud Native API Management
The Kubernetes Gateway API and its role in Cloud Native API Management
 
Computer 10: Lesson 10 - Online Crimes and Hazards
Computer 10: Lesson 10 - Online Crimes and HazardsComputer 10: Lesson 10 - Online Crimes and Hazards
Computer 10: Lesson 10 - Online Crimes and Hazards
 
201610817 - edge part1
201610817 - edge part1201610817 - edge part1
201610817 - edge part1
 
COMPUTER 10: Lesson 7 - File Storage and Online Collaboration
COMPUTER 10: Lesson 7 - File Storage and Online CollaborationCOMPUTER 10: Lesson 7 - File Storage and Online Collaboration
COMPUTER 10: Lesson 7 - File Storage and Online Collaboration
 
Using IESVE for Loads, Sizing and Heat Pump Modeling to Achieve Decarbonization
Using IESVE for Loads, Sizing and Heat Pump Modeling to Achieve DecarbonizationUsing IESVE for Loads, Sizing and Heat Pump Modeling to Achieve Decarbonization
Using IESVE for Loads, Sizing and Heat Pump Modeling to Achieve Decarbonization
 
Anypoint Code Builder , Google Pub sub connector and MuleSoft RPA
Anypoint Code Builder , Google Pub sub connector and MuleSoft RPAAnypoint Code Builder , Google Pub sub connector and MuleSoft RPA
Anypoint Code Builder , Google Pub sub connector and MuleSoft RPA
 
IEEE Computer Society’s Strategic Activities and Products including SWEBOK Guide
IEEE Computer Society’s Strategic Activities and Products including SWEBOK GuideIEEE Computer Society’s Strategic Activities and Products including SWEBOK Guide
IEEE Computer Society’s Strategic Activities and Products including SWEBOK Guide
 
UiPath Community: AI for UiPath Automation Developers
UiPath Community: AI for UiPath Automation DevelopersUiPath Community: AI for UiPath Automation Developers
UiPath Community: AI for UiPath Automation Developers
 

Social Aspects of Interactive Recommender Systems

  • 1. Social aspects of interactive RecSys: Bridging the gap between predictive algorithms and interactive user interfaces Denis Parra, Assistant Professor CS Department School of Engineering Pontificia Universidad Católica deChile SoAPS Workshop at ECIR 2018, March 26th 2018 Funded by :
  • 2. Personal Introduction • 2008-2013: PhD at U. of Pittsburgh • 2013 – now: CS Department, PUC (Santiago, Chile) March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 2 8,280 Km
  • 3. Outline • Quick intro to – RecommenderSystems – Visualization in RecommenderSystems • Survey of our work on Visualization and Interaction on RecSys – With highlights to social aspects of the results • Summary & Discussion March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 3
  • 4. In Collaboration with • Peter Brusilovsky (University of Pittsburgh, USA) • Katrien Verbert (KU Leuven, Belgium) • Christoph Trattner (University of Bergen, Norway) • Chaoli Wang (Notre Dame University, USA) • Ivania Donoso (alumni, PUC Chile) • María Sepúlveda (MSc student, PUC Chile) March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 4
  • 5. INTRODUCTIONTORECSYS Recommender Systems * Danboard (Danbo):Amazon’s cardboard robot, in these slides it represents a recommender system *
  • 6. Recommender Systems (RecSys) Systems that help (groups of) people to find relevant items in a crowdeditem or information space(MacNee et al. 2006) March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 6
  • 7. Why do we care about RecSys? • Nowadays, several domains& applications require people to make decisions among a large set of items. March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 7
  • 8. A lil’ bit of History • First recommender systems were built at the beginning of 90’s (Tapestry, GroupLens, Ringo) • Online contests, such as the Netflix prize, grew the attention on recommender systems beyond Computer Science (2006-2009) March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 8
  • 9. The Recommendation Problem • A popular way of presenting the recommendation problem was rating prediction (Netflix prize) • How good is my prediction? Item 1 Item 2 … Item m User 1 1 5 4 User 2 5 1 ? … User n 2 5 ? Predict! March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 9
  • 10. Traditional Recommendation Methods • Without covering all possible methods, the two most typical classifications on recommender algorithms are Classification 1 Classification 2 - Collaborative Filtering - Content-based Filtering - Hybrid - Memory-based - Model-based March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 10
  • 11. Collaborative Filtering (User-based KNN) • Find like-minded people to recommend 5 4 4 2 1 5 4 4 Active user User_1 User_2 ∑ ∑ ⊂ ⊂ −⋅ += )( )( ),( )(),( ),( uneighborsn uneighborsn nni u nuuserSim rrnuuserSim riupred2 3 4 2 Item 1 Item 2 Item 3 March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 11
  • 12. Content-Based Filtering • Can be traced back to techniques from IR, where the User Profile represents a query. user_profile = {w_1, w_2, …., w_3} using TF-IDF, weighting Doc_1 = {w_1, w_2, …., w_3} Doc_2 = {w_1, w_2, …., w_3} Doc_3 = {w_1, w_2, …., w_3} Doc_n = {w_1, w_2, …., w_3} 5 4 5 March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 12
  • 13. Hybridization • Combine previous methods to overcome their weaknesses (Burke, 2002) March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 13
  • 14. Model-based: Matrix Factorization Latent vector of the item Latent vector of the user SVD ~ Singular Value Decomposition March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 14
  • 15. Other paradigms and techniques • Recommendation as a graph problem: – PersonalizedPageRank (Kamvar et al, 2010), (Santos et al 2016), etc. • Recommendation as a ranking problem: – Karatzoglou et al. (2013), Shi et al. (2014), Macedo et al. (2015), etc. • Deep learning methods: – MetaProd2Vecby Vasile et al. (2016), YouTube Recommendationsby Covington et al. (2016), etc. March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 15
  • 16. Are accuracy metrics the goal of RecSys? • Some works started pointing out that small improvements in RMSE did not have a proportional improvement in user satisfaction March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 16
  • 17. MovieLens: Traditional RecSys Interface March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 17
  • 18. PeerChooser (2008) Controllability in CF March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 18 O’Donovan et al. “PeerChooser: Visual Interactive Recommendation” (2008)
  • 19. SmallWorlds: Expanded Explainability March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 19 Gretarsson et al. “SmallWorlds:Visualizing social recommendations” (2010)
  • 20. TasteWeights: Hybrid Control and Inspect Bostandjev et al. “TasteWeights:A Visual Interactive Hybrid Recommender System” (2012) Controllability: Sliders that let users control the importance of preferences and contexts Inspectability: lines that connect recommended items with contexts and user preferences March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 20
  • 21. IUI 2017 • Loepp et al. (2017) March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 21
  • 22. More Details? Check our survey March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 22 He, C., Parra, D., & Verbert, K. (2016). Interactive recommender systems: a survey of the state of the art and future research challenges andopportunities.Expert Systems with Applications, 56, 9-27.
  • 23. My Take on RecSys Research (2009 ~) March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 23
  • 24. IUI 2013 IUI 2014 UMAP 2016 E.I.-VDA 2018 IUI 2018 Visual & interactive RecSys interfaces March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 24 CNVis EpistAid TalkExplorer SetFusion Moodplay
  • 25. March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 25
  • 26. TalkExplorer – IUI 2013 • Adaptation of Aduna Visualization to CN • Main researchquestion: Does fusion (intersection) of contextsof relevance improve user experience with RecSys? March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 26
  • 27. Research Platform • The studies were conducted using Conference Navigator, a Conference Support System • Our goal was recommending conference talks Program Proceedings Author List Recommendations http://halley.exp.sis.pitt.edu/cn3/ March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 27
  • 28. TalkExplorer - I Entities Tags, Recommender Agents, Users March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 28
  • 29. TalkExplorer - II Recommender Recommender Cluster with intersecti on of entities Cluster (of talks) associated to only one entity • Canvas Area: Intersections of Different Entities User March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 29
  • 30. TalkExplorer - III Items Talks explored by the user March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 30
  • 31. Evaluation: Intersections & Effectiveness • What do we call an “Intersection”? • We used #explorations on intersections and their effectiveness, defined as: Effectiveness = March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 31
  • 32. 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 March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 32
  • 33. TalkExplorer Studies I & II • Study I – ControlledExperiment:Userswere asked todiscover relevant talksby exploringthe three types of entities:tags, recommenderagents and users. – Conductedat HypertextandUMAP 2012(21 users) – Subjectsfamiliarwith Visualizationsand Recsys • Study II – FieldStudy: Users were left free to explore the interface. – Conductedat LAK 2012 and ECTEL 2013 (18 users) – Subjectsfamiliarwith visualizations, but not much with RecSys March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 33
  • 34. Results of Studies I & II • Effectiveness increases with intersections of more entities • Effectiveness wasn’t affected in the field study (study 2) • … but exploration distribution was affected March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 34
  • 35. More detail on entities March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 35
  • 36. More detail on entities March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 36 • Despite not being the most effective, the user (social) entity attracted by far more explorations from users.
  • 37. Social ~ Trust in RecSys • O'Donovan, John, and Barry Smyth. "Trust in recommender systems," IUI, 2005. • Golbeck, Jennifer, and James Hendler. "Filmtrust: Movie recommendations using trust in web-based social networks," ComNet, 2006. • Jamali, Mohsen, and Martin Ester. "A matrix factorization technique with trust propagation for recommendation in social networks." RecSys, 2010. March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 37
  • 38. SetFusion – IUI 2014 March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 38
  • 39. SetFusion I Traditional Ranked List Papers sorted by Relevance. It combines 3 recommendation approaches. March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 39
  • 40. SetFusion - II Sliders Allow the user to control the importance of each data source or recommendation method Interactive Venn Diagram Allows the user to inspect and to filter papers recommended. Actions available: - Filter item list by clicking on an area - Highlight a paper by mouse-over on a circle - Scroll to paper by clicking on a circle - Indicate bookmarked papers March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 40
  • 41. Study : iConference March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 41 A B C • A and C: Social • B: Content-based
  • 42. Rating per method – Effect of Visuals March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 42
  • 43. Rating per method – Effect of Visuals March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 43 No significant difference
  • 44. Rating per method – Effect of Visuals March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 44 Hybrid algorithm + visualization yield the only significant difference
  • 45. Summary & Conclusions • The combination of several sources of relevance has an impact on recommendation, being the social aspect among the most relevant. • The visual paradigm combined with social aspects used can have a significant effect on user behavior. • We need to keep working on visual representations that promote exploration without decreasing recommendation accuracy. March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 45
  • 46. MOODFROM SOCIAL SOURCES FOR MUSICRECOMMENDATION
  • 47. Moodplay • MoodPlay – With Ivana Andjelkovic& John O’Donovan (UCSB) March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 47 Andjelkovic,I., Parra, D., & O'Donovan,J. (2016). Moodplay:Interactive Mood-based Music Discovery and Recommendation.In Proceedings of the 2016 Conference on User Modeling Adaptation and Personalization (pp. 275-279).ACM.
  • 48. MoodPlay March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 48
  • 49. System Architecture March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 49
  • 50. Emotion Models • Modelo de emociones de Russel (1980) • GEMS (2008) March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 50
  • 51. Moods and Music: the GEMS model March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 51
  • 52. Components • Colors: GEMS emotions • Dots: artists • Zooming and Panning • Playing & Recommendations March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 52
  • 53. Components • Colors: GEMS emotions • Dots: artists • Zooming and Panning • Playing & Recommendations March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 53
  • 54. Components • Colors: GEMS emotions • Dots: artists • Zooming and Panning • Playing & Recommendations March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 54
  • 55. Components • Colors: GEMS emotions • Dots: artists • Zooming and Panning • Playing & Recommendations March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 55
  • 56. Emotion-Aware Recommendation • Using several Web APIs, we collected users’ perception of mood associated to artists. • Then, using artists as input, we calculate recommendations of similarly perceived artists March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 56
  • 57. MoodPlay March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 57 • Our results indicate that users’ reported mood and artist mood have an effect on people satisfaction with the system. • Try it at http://moodplay.pythonanywhere.com
  • 59. March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 59 Interactive Relevance Feedback Interface for evidence–based Medicine Ivania Donoso-Guzmán Denis Parra Honorable mention for best paper award ACM IUI 2018
  • 60. Belleret al (2013). … takes from 6 to 12 meses Process for answering a clinical question 60 March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018)
  • 61. 61 March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) EpistAid
  • 62. Model Interfaz N Documentos Vistos Ground Truth Recall Precision F-1 score Recall BM25 Non-Viz 12 0.66+/-0.08 0.52 +/- 0.06 0.58 +/- 0.07 0.20 +/- 0.04 BM25 Viz 11 0.71+/-0.06 0.64 +/- 0.04 0.64 +/- 0.03 0.18 +/- 0.02 Rocchio Non-Viz 11 0.65+/- 0.08 0.73 +/- 0.02 0.65 +/- 0.05 0.21 +/- 0.04 Rocchio Viz 12 0.77 +/- 0.06 0.67 +/- 0.01 0.70 +/- 0.03 0.23 +/- 0.03 Promising Results 62 March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018)
  • 63. Future Work March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 63 - CollaborativeModel: - Whichis the best way to providea collaborative human-in-the-loopinterfacefor Evidence-Based Medicine? - … Ideas from research on collaborative web search, for instance: Yue et al. “Influences on query reformulation in collaborative web search”. IEEE Computer Magazine, 2014.
  • 64. IEEE VIS 2017 – Panel ML & Vis ? March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 64
  • 65. Deep Learning and Social on RecSys • Lei, Chenyi, et al. "Comparative deep learning of hybrid representations for image recommendations." CVPR. 2016. March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 65
  • 66. Deep Learning and Social on RecSys • Lei, Chenyi, et al. "Comparative deep learning of hybrid representations for image recommendations." CVPR. 2016. March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 66 - How to make better sense of the social embeddings for explaining recommendations?
  • 67. IEEE VIS 2017 – Panel ML & Vis ? March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 67
  • 68. Conclusion March 26th, 2018 D.Parra~ Invited talk at SoAPSWS(ECIR 2018) 68 • The social aspects are indeed relevant in recommendation systems, and their effect on prediction accuracyhas been already studied. • The studies presented in this talk show that combining interactive visualizations with social and other relevance signal can have an important effect on users’ perception of recommendations. • I invite researcher to further study the connections between social aspects, visualization and their effect on recommender systems.