SlideShare a Scribd company logo
1 of 17
Unifying Explicit and Implicit Feedback for
Rating Prediction and Ranking
Recommendation Tasks
Amir H. Jadidinejad, Craig Macdonald, Iadh Ounis
University of Glasgow
ICTIR 2019
Introduction
When users visit recommendation websites
they generate two types of feedback
• Explicit Feedback: The user is consciously
providing information to the system
• Implicit Feedback: The system is collecting
feedback based on users behaviour (invisibly
to them)
These two types of feedback can be used to
automate recommendation tasks
• Explicit → Rating Prediction (e.g. MF)
• Implicit → Ranking (e.g. BPR)
2
Rating Reviews Likes
Clicks
Motivation
3
M. Wan and J. McAuley. Item Recommendation on Monotonic Behavior Chains, RecSys 2018.
We also aim to unify explicit and implicit feedback in order to
bridge the rating prediction and ranking prediction tasks
click purchase review recommend
feedback type
Our Approach wrt Existing
Literature
Previous unifying approaches tackled the problem at the level of
recommendation model
• i.e. they proposed new models to combine both explicit and implicit
feedback
The main barrier in unifying explicit and implicit feedback is that
they are heterogeneous in terms of:
• Representations: <u,i,r> vs <u,i>
• Distributions: 90% clicks, 10% reviews
4
Instead, we propose a novel weak supervision approach that bridges the gap
between explicit and implicit feedback at the level of data preprocessing.
Previous Unifying Approaches
Models that leverage a shared representation of users and items
• Joint Representation Learning for Top-N Recommendation with
Heterogeneous Information Sources (Zhang et al. CIKM 2017)
• Personalized Ranking Recommendation via Integrating Multiple Feedbacks
(Liu et al. PAKDD 2017)
Models that extend existing recommendation algorithms
• Item Recommendation on Monotonic Behavior Chains. In Proceedings of
(Wan and McAuley. RecSys 2018)
• Bayesian Personalized Ranking with Multi-Channel User Feedback (Loni et
al. RecSys 2016)
5
Our Proposed Approach (1)
6
Our Proposal: Transfer the knowledge of rating prediction
from explicit feedback to annotate implicit feedback
𝐷𝑖 = ⟨𝑈, 𝐼⟩
Idea: Can we use explicit feedback De to infer the user ratings of Di?
𝐷𝑖
∗
= ⟨𝑈, 𝐼, 𝑅∗⟩
Our Proposed Approach (2)
1. Train a model based on explicit
feedback (ratings) :
7
Ratings
Clicks
Our Proposed Approach (2)
1. Train a model based on explicit
feedback (ratings) :
2. Use the model from (1) to annotate
implicit (click) feedback by predicting
the rating values -> :
3. Train a new model based on the
annotated dataset :
8
Ratings
Predicted
ratings
for clicks
Features of our Unifying
Approach
Approach
Rating
Prediction
Ranking Long-tail
Coverage
Explicit Models
(MF)
Implicit Models
(BPR)
Our Approach
9
Our aim is to improve the capability of the explicit models for the task of ranking
through the use of a novel weak supervision approach that unifies both explicit
and implicit feedback datasets
Research Questions
• RQ1: Does the proposed weak supervision approach
enhance recommendation accuracy for
(a) Rating Prediction
(b) Ranking
• RQ2: How much popularity bias is exhibited by our
proposed weak supervision approach compared with
the other baseline recommendation approaches?
10
Datasets
11
• GoodReads: Large book ratings dataset
• Steam: Video game reviews
• Douban: Chinese movie, book and music ratings
• Dianping: User ratings from DianPing.com
Evaluation Metrics
RQ1:
• Rating Prediction:
- Root Mean Square Error (RMSE)
• Ranking:
- MRR, nDCG, MAP
RQ2:
• Long-tail item coverage:
- Average Recommendation Popularity
(ARP) by Yin et al. (VLDB12):
12
Reviews
80/10/10
Clicks
Baselines
1. A popularity model
2. An explicit model
- Matrix Factorization:
3. An implicit model
- Bayesian Personalized Ranking:
4. An existing unifying model
- ChainRec (Wan and McAuley, RecSys 2018)
13
14
RQ1.a
Φ 𝐷 𝑒
~ ΦDi
∗
RQ1.b
ΦDi
∗
>> Φ 𝐷 𝑒
𝑀𝐹𝐵𝑃𝑅 > ΦDi
∗
RQ2
ΦDi
∗
>> 𝑀𝐹𝐵𝑃𝑅
The proposed weak
supervision approach
improves the capability of
the explicit models for the
task of ranking (5 out of 6
datasets) and has a
better coverage of tail
items than BPR.
Qualitative Analysis
15
Over 70% of BPR’s recommended
items are among the top-20 most
popular items
Only 15% of the explicit
recommendations are among the top-20
most popular items (long tail coverage)
Conclusions and Future Work
Our approach unifies both the Rating prediction (explicit feedback)
and Ranking (implicit feedback) tasks while alleviating the bias
against less popular long-tail items.
• It is applied at the level of data pre-processing as a weak supervision
signal.
Future Work
• Combining weak supervision into pairwise approaches such as BPR or
neural network-based recommenders
• Tradeoff between good quality recommendations and long-tail item
coverage requires more investigation.
16
17
Thanks!
• Data and code is available at:
- https://github.com/amirj/unifying_explicit_implicit

More Related Content

What's hot

[WI 2017] Context Suggestion: Empirical Evaluations vs User Studies
[WI 2017] Context Suggestion: Empirical Evaluations vs User Studies[WI 2017] Context Suggestion: Empirical Evaluations vs User Studies
[WI 2017] Context Suggestion: Empirical Evaluations vs User StudiesYONG ZHENG
 
Trust and Recommender Systems
Trust and  Recommender SystemsTrust and  Recommender Systems
Trust and Recommender Systemszhayefei
 
AIRG Presentation
AIRG PresentationAIRG Presentation
AIRG Presentationnirvdrum
 
Text Enhanced Recommendation System Model Based on Yelp Reviews
Text Enhanced Recommendation System Model Based on Yelp ReviewsText Enhanced Recommendation System Model Based on Yelp Reviews
Text Enhanced Recommendation System Model Based on Yelp ReviewsHari Sanadhya
 
Collaborative Metric Learning (WWW'17)
Collaborative Metric Learning (WWW'17)Collaborative Metric Learning (WWW'17)
Collaborative Metric Learning (WWW'17)承剛 謝
 
Starke2017 - Effective User Interface Designs to Increase Energy-efficient Be...
Starke2017 - Effective User Interface Designs to Increase Energy-efficient Be...Starke2017 - Effective User Interface Designs to Increase Energy-efficient Be...
Starke2017 - Effective User Interface Designs to Increase Energy-efficient Be...Alain Starke
 
Thesis Presentation
Thesis PresentationThesis Presentation
Thesis Presentationnirvdrum
 
Temporal Learning and Sequence Modeling for a Job Recommender System
Temporal Learning and Sequence Modeling for a Job Recommender SystemTemporal Learning and Sequence Modeling for a Job Recommender System
Temporal Learning and Sequence Modeling for a Job Recommender SystemAnoop Kumar
 
Proceedings Template - WORD
Proceedings Template - WORDProceedings Template - WORD
Proceedings Template - WORDbutest
 
A Combination of Simple Models by Forward Predictor Selection for Job Recomme...
A Combination of Simple Models by Forward Predictor Selection for Job Recomme...A Combination of Simple Models by Forward Predictor Selection for Job Recomme...
A Combination of Simple Models by Forward Predictor Selection for Job Recomme...David Zibriczky
 
Asking Clarifying Questions in Open-Domain Information-Seeking Conversations
Asking Clarifying Questions in Open-Domain Information-Seeking ConversationsAsking Clarifying Questions in Open-Domain Information-Seeking Conversations
Asking Clarifying Questions in Open-Domain Information-Seeking ConversationsMohammad Aliannejadi
 
Collaborative personalized tweet recommendation
Collaborative personalized tweet recommendationCollaborative personalized tweet recommendation
Collaborative personalized tweet recommendationLiangjie Hong
 
Recsys2021_slides_sato
Recsys2021_slides_satoRecsys2021_slides_sato
Recsys2021_slides_satoMasahiro Sato
 
Pedagogical and Technological Interventions in Peer Review (Shivers-McNair) #...
Pedagogical and Technological Interventions in Peer Review (Shivers-McNair) #...Pedagogical and Technological Interventions in Peer Review (Shivers-McNair) #...
Pedagogical and Technological Interventions in Peer Review (Shivers-McNair) #...annshiversmcnair
 
Collaborative Filtering 1: User-based CF
Collaborative Filtering 1: User-based CFCollaborative Filtering 1: User-based CF
Collaborative Filtering 1: User-based CFYusuke Yamamoto
 
Query/Task Satisfaction and Grid-based Evaluation Metrics Under Different Ima...
Query/Task Satisfaction and Grid-based Evaluation Metrics Under Different Ima...Query/Task Satisfaction and Grid-based Evaluation Metrics Under Different Ima...
Query/Task Satisfaction and Grid-based Evaluation Metrics Under Different Ima...Kosetsu Tsukuda
 
Harm van Seijen, Research Scientist, Maluuba at MLconf SF 2016
Harm van Seijen, Research Scientist, Maluuba at MLconf SF 2016Harm van Seijen, Research Scientist, Maluuba at MLconf SF 2016
Harm van Seijen, Research Scientist, Maluuba at MLconf SF 2016MLconf
 
Doctoral Consortium Slides at SIGIR 2017
Doctoral Consortium Slides at SIGIR 2017Doctoral Consortium Slides at SIGIR 2017
Doctoral Consortium Slides at SIGIR 2017Jarana Manotumruksa
 
Opinion dynamics(opinion dynamics based group recommender systems) screen...
Opinion dynamics(opinion dynamics based group recommender systems)     screen...Opinion dynamics(opinion dynamics based group recommender systems)     screen...
Opinion dynamics(opinion dynamics based group recommender systems) screen...Venkat Projects
 

What's hot (20)

[WI 2017] Context Suggestion: Empirical Evaluations vs User Studies
[WI 2017] Context Suggestion: Empirical Evaluations vs User Studies[WI 2017] Context Suggestion: Empirical Evaluations vs User Studies
[WI 2017] Context Suggestion: Empirical Evaluations vs User Studies
 
Trust and Recommender Systems
Trust and  Recommender SystemsTrust and  Recommender Systems
Trust and Recommender Systems
 
AIRG Presentation
AIRG PresentationAIRG Presentation
AIRG Presentation
 
Text Enhanced Recommendation System Model Based on Yelp Reviews
Text Enhanced Recommendation System Model Based on Yelp ReviewsText Enhanced Recommendation System Model Based on Yelp Reviews
Text Enhanced Recommendation System Model Based on Yelp Reviews
 
Collaborative Metric Learning (WWW'17)
Collaborative Metric Learning (WWW'17)Collaborative Metric Learning (WWW'17)
Collaborative Metric Learning (WWW'17)
 
Starke2017 - Effective User Interface Designs to Increase Energy-efficient Be...
Starke2017 - Effective User Interface Designs to Increase Energy-efficient Be...Starke2017 - Effective User Interface Designs to Increase Energy-efficient Be...
Starke2017 - Effective User Interface Designs to Increase Energy-efficient Be...
 
Thesis Presentation
Thesis PresentationThesis Presentation
Thesis Presentation
 
Recommender Systems
Recommender SystemsRecommender Systems
Recommender Systems
 
Temporal Learning and Sequence Modeling for a Job Recommender System
Temporal Learning and Sequence Modeling for a Job Recommender SystemTemporal Learning and Sequence Modeling for a Job Recommender System
Temporal Learning and Sequence Modeling for a Job Recommender System
 
Proceedings Template - WORD
Proceedings Template - WORDProceedings Template - WORD
Proceedings Template - WORD
 
A Combination of Simple Models by Forward Predictor Selection for Job Recomme...
A Combination of Simple Models by Forward Predictor Selection for Job Recomme...A Combination of Simple Models by Forward Predictor Selection for Job Recomme...
A Combination of Simple Models by Forward Predictor Selection for Job Recomme...
 
Asking Clarifying Questions in Open-Domain Information-Seeking Conversations
Asking Clarifying Questions in Open-Domain Information-Seeking ConversationsAsking Clarifying Questions in Open-Domain Information-Seeking Conversations
Asking Clarifying Questions in Open-Domain Information-Seeking Conversations
 
Collaborative personalized tweet recommendation
Collaborative personalized tweet recommendationCollaborative personalized tweet recommendation
Collaborative personalized tweet recommendation
 
Recsys2021_slides_sato
Recsys2021_slides_satoRecsys2021_slides_sato
Recsys2021_slides_sato
 
Pedagogical and Technological Interventions in Peer Review (Shivers-McNair) #...
Pedagogical and Technological Interventions in Peer Review (Shivers-McNair) #...Pedagogical and Technological Interventions in Peer Review (Shivers-McNair) #...
Pedagogical and Technological Interventions in Peer Review (Shivers-McNair) #...
 
Collaborative Filtering 1: User-based CF
Collaborative Filtering 1: User-based CFCollaborative Filtering 1: User-based CF
Collaborative Filtering 1: User-based CF
 
Query/Task Satisfaction and Grid-based Evaluation Metrics Under Different Ima...
Query/Task Satisfaction and Grid-based Evaluation Metrics Under Different Ima...Query/Task Satisfaction and Grid-based Evaluation Metrics Under Different Ima...
Query/Task Satisfaction and Grid-based Evaluation Metrics Under Different Ima...
 
Harm van Seijen, Research Scientist, Maluuba at MLconf SF 2016
Harm van Seijen, Research Scientist, Maluuba at MLconf SF 2016Harm van Seijen, Research Scientist, Maluuba at MLconf SF 2016
Harm van Seijen, Research Scientist, Maluuba at MLconf SF 2016
 
Doctoral Consortium Slides at SIGIR 2017
Doctoral Consortium Slides at SIGIR 2017Doctoral Consortium Slides at SIGIR 2017
Doctoral Consortium Slides at SIGIR 2017
 
Opinion dynamics(opinion dynamics based group recommender systems) screen...
Opinion dynamics(opinion dynamics based group recommender systems)     screen...Opinion dynamics(opinion dynamics based group recommender systems)     screen...
Opinion dynamics(opinion dynamics based group recommender systems) screen...
 

Similar to ACM ICTIR 2019 Slides - Santa Clara, USA

Preference Elicitation Interface
Preference Elicitation InterfacePreference Elicitation Interface
Preference Elicitation Interface晓愚 孟
 
IRJET- Analysis of Question and Answering Recommendation System
IRJET-  	  Analysis of Question and Answering Recommendation SystemIRJET-  	  Analysis of Question and Answering Recommendation System
IRJET- Analysis of Question and Answering Recommendation SystemIRJET Journal
 
Rokach-GomaxSlides.pptx
Rokach-GomaxSlides.pptxRokach-GomaxSlides.pptx
Rokach-GomaxSlides.pptxJadna Almeida
 
Rokach-GomaxSlides (1).pptx
Rokach-GomaxSlides (1).pptxRokach-GomaxSlides (1).pptx
Rokach-GomaxSlides (1).pptxJadna Almeida
 
[IUI 2017] Criteria Chains: A Novel Multi-Criteria Recommendation Approach
[IUI 2017] Criteria Chains: A Novel Multi-Criteria Recommendation Approach[IUI 2017] Criteria Chains: A Novel Multi-Criteria Recommendation Approach
[IUI 2017] Criteria Chains: A Novel Multi-Criteria Recommendation ApproachYONG ZHENG
 
A Review of Socialized Word Embeddings (Zeng+, 2017)
A Review of Socialized Word Embeddings (Zeng+, 2017)A Review of Socialized Word Embeddings (Zeng+, 2017)
A Review of Socialized Word Embeddings (Zeng+, 2017)Shuntaro Yada
 
HABIB FIGA GUYE {BULE HORA UNIVERSITY}(habibifiga@gmail.com
HABIB FIGA GUYE {BULE HORA UNIVERSITY}(habibifiga@gmail.comHABIB FIGA GUYE {BULE HORA UNIVERSITY}(habibifiga@gmail.com
HABIB FIGA GUYE {BULE HORA UNIVERSITY}(habibifiga@gmail.comHABIB FIGA GUYE
 
Summarization and opinion detection of product reviews (1)
Summarization and opinion detection of product reviews (1)Summarization and opinion detection of product reviews (1)
Summarization and opinion detection of product reviews (1)Lokesh Mittal
 
Social Recommender Systems Tutorial - WWW 2011
Social Recommender Systems Tutorial - WWW 2011Social Recommender Systems Tutorial - WWW 2011
Social Recommender Systems Tutorial - WWW 2011idoguy
 
DaCENA Personalized Exploration of Knowledge Graphs Within a Context. Seminar...
DaCENA Personalized Exploration of Knowledge Graphs Within a Context. Seminar...DaCENA Personalized Exploration of Knowledge Graphs Within a Context. Seminar...
DaCENA Personalized Exploration of Knowledge Graphs Within a Context. Seminar...Università degli Studi di Milano-Bicocca
 
Best Practices in Recommender System Challenges
Best Practices in Recommender System ChallengesBest Practices in Recommender System Challenges
Best Practices in Recommender System ChallengesAlan Said
 
Which Vertical Search Engines are Relevant? Understanding Vertical Relevance ...
Which Vertical Search Engines are Relevant? Understanding Vertical Relevance ...Which Vertical Search Engines are Relevant? Understanding Vertical Relevance ...
Which Vertical Search Engines are Relevant? Understanding Vertical Relevance ...Mounia Lalmas-Roelleke
 
Forecasting movie rating using k-nearest neighbor based collaborative filtering
Forecasting movie rating using k-nearest neighbor based  collaborative filteringForecasting movie rating using k-nearest neighbor based  collaborative filtering
Forecasting movie rating using k-nearest neighbor based collaborative filteringIJECEIAES
 
V Jornadas eMadrid sobre “Educación Digital”. Roberto Centeno, Universidad Na...
V Jornadas eMadrid sobre “Educación Digital”. Roberto Centeno, Universidad Na...V Jornadas eMadrid sobre “Educación Digital”. Roberto Centeno, Universidad Na...
V Jornadas eMadrid sobre “Educación Digital”. Roberto Centeno, Universidad Na...eMadrid network
 
Experiments on Generalizability of User-Oriented Fairness in Recommender Systems
Experiments on Generalizability of User-Oriented Fairness in Recommender SystemsExperiments on Generalizability of User-Oriented Fairness in Recommender Systems
Experiments on Generalizability of User-Oriented Fairness in Recommender SystemsHossein A. (Saeed) Rahmani
 
Collaborative Filtering Recommendation System
Collaborative Filtering Recommendation SystemCollaborative Filtering Recommendation System
Collaborative Filtering Recommendation SystemMilind Gokhale
 
ICSME 2016: Search-Based Peer Reviewers Recommendation in Modern Code Review
ICSME 2016: Search-Based Peer Reviewers Recommendation in Modern Code ReviewICSME 2016: Search-Based Peer Reviewers Recommendation in Modern Code Review
ICSME 2016: Search-Based Peer Reviewers Recommendation in Modern Code ReviewAli Ouni
 
Enhancing Multi-Aspect Collaborative Filtering for Personalized Recommendation
Enhancing Multi-Aspect Collaborative Filtering for Personalized RecommendationEnhancing Multi-Aspect Collaborative Filtering for Personalized Recommendation
Enhancing Multi-Aspect Collaborative Filtering for Personalized RecommendationNurfadhlina Mohd Sharef
 
Crowdsourcing Predictors of Behavioral Outcomes
Crowdsourcing Predictors of Behavioral OutcomesCrowdsourcing Predictors of Behavioral Outcomes
Crowdsourcing Predictors of Behavioral OutcomesAlekya Yermal
 
Finding Co-solvers on Twitter, with the Little Help from Linked Data
Finding Co-solvers on Twitter, with the Little Help from Linked DataFinding Co-solvers on Twitter, with the Little Help from Linked Data
Finding Co-solvers on Twitter, with the Little Help from Linked DataMilan Stankovic
 

Similar to ACM ICTIR 2019 Slides - Santa Clara, USA (20)

Preference Elicitation Interface
Preference Elicitation InterfacePreference Elicitation Interface
Preference Elicitation Interface
 
IRJET- Analysis of Question and Answering Recommendation System
IRJET-  	  Analysis of Question and Answering Recommendation SystemIRJET-  	  Analysis of Question and Answering Recommendation System
IRJET- Analysis of Question and Answering Recommendation System
 
Rokach-GomaxSlides.pptx
Rokach-GomaxSlides.pptxRokach-GomaxSlides.pptx
Rokach-GomaxSlides.pptx
 
Rokach-GomaxSlides (1).pptx
Rokach-GomaxSlides (1).pptxRokach-GomaxSlides (1).pptx
Rokach-GomaxSlides (1).pptx
 
[IUI 2017] Criteria Chains: A Novel Multi-Criteria Recommendation Approach
[IUI 2017] Criteria Chains: A Novel Multi-Criteria Recommendation Approach[IUI 2017] Criteria Chains: A Novel Multi-Criteria Recommendation Approach
[IUI 2017] Criteria Chains: A Novel Multi-Criteria Recommendation Approach
 
A Review of Socialized Word Embeddings (Zeng+, 2017)
A Review of Socialized Word Embeddings (Zeng+, 2017)A Review of Socialized Word Embeddings (Zeng+, 2017)
A Review of Socialized Word Embeddings (Zeng+, 2017)
 
HABIB FIGA GUYE {BULE HORA UNIVERSITY}(habibifiga@gmail.com
HABIB FIGA GUYE {BULE HORA UNIVERSITY}(habibifiga@gmail.comHABIB FIGA GUYE {BULE HORA UNIVERSITY}(habibifiga@gmail.com
HABIB FIGA GUYE {BULE HORA UNIVERSITY}(habibifiga@gmail.com
 
Summarization and opinion detection of product reviews (1)
Summarization and opinion detection of product reviews (1)Summarization and opinion detection of product reviews (1)
Summarization and opinion detection of product reviews (1)
 
Social Recommender Systems Tutorial - WWW 2011
Social Recommender Systems Tutorial - WWW 2011Social Recommender Systems Tutorial - WWW 2011
Social Recommender Systems Tutorial - WWW 2011
 
DaCENA Personalized Exploration of Knowledge Graphs Within a Context. Seminar...
DaCENA Personalized Exploration of Knowledge Graphs Within a Context. Seminar...DaCENA Personalized Exploration of Knowledge Graphs Within a Context. Seminar...
DaCENA Personalized Exploration of Knowledge Graphs Within a Context. Seminar...
 
Best Practices in Recommender System Challenges
Best Practices in Recommender System ChallengesBest Practices in Recommender System Challenges
Best Practices in Recommender System Challenges
 
Which Vertical Search Engines are Relevant? Understanding Vertical Relevance ...
Which Vertical Search Engines are Relevant? Understanding Vertical Relevance ...Which Vertical Search Engines are Relevant? Understanding Vertical Relevance ...
Which Vertical Search Engines are Relevant? Understanding Vertical Relevance ...
 
Forecasting movie rating using k-nearest neighbor based collaborative filtering
Forecasting movie rating using k-nearest neighbor based  collaborative filteringForecasting movie rating using k-nearest neighbor based  collaborative filtering
Forecasting movie rating using k-nearest neighbor based collaborative filtering
 
V Jornadas eMadrid sobre “Educación Digital”. Roberto Centeno, Universidad Na...
V Jornadas eMadrid sobre “Educación Digital”. Roberto Centeno, Universidad Na...V Jornadas eMadrid sobre “Educación Digital”. Roberto Centeno, Universidad Na...
V Jornadas eMadrid sobre “Educación Digital”. Roberto Centeno, Universidad Na...
 
Experiments on Generalizability of User-Oriented Fairness in Recommender Systems
Experiments on Generalizability of User-Oriented Fairness in Recommender SystemsExperiments on Generalizability of User-Oriented Fairness in Recommender Systems
Experiments on Generalizability of User-Oriented Fairness in Recommender Systems
 
Collaborative Filtering Recommendation System
Collaborative Filtering Recommendation SystemCollaborative Filtering Recommendation System
Collaborative Filtering Recommendation System
 
ICSME 2016: Search-Based Peer Reviewers Recommendation in Modern Code Review
ICSME 2016: Search-Based Peer Reviewers Recommendation in Modern Code ReviewICSME 2016: Search-Based Peer Reviewers Recommendation in Modern Code Review
ICSME 2016: Search-Based Peer Reviewers Recommendation in Modern Code Review
 
Enhancing Multi-Aspect Collaborative Filtering for Personalized Recommendation
Enhancing Multi-Aspect Collaborative Filtering for Personalized RecommendationEnhancing Multi-Aspect Collaborative Filtering for Personalized Recommendation
Enhancing Multi-Aspect Collaborative Filtering for Personalized Recommendation
 
Crowdsourcing Predictors of Behavioral Outcomes
Crowdsourcing Predictors of Behavioral OutcomesCrowdsourcing Predictors of Behavioral Outcomes
Crowdsourcing Predictors of Behavioral Outcomes
 
Finding Co-solvers on Twitter, with the Little Help from Linked Data
Finding Co-solvers on Twitter, with the Little Help from Linked DataFinding Co-solvers on Twitter, with the Little Help from Linked Data
Finding Co-solvers on Twitter, with the Little Help from Linked Data
 

Recently uploaded

Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024BookNet Canada
 
Unblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesUnblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesSinan KOZAK
 
Connect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck PresentationConnect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck PresentationSlibray Presentation
 
Streamlining Python Development: A Guide to a Modern Project Setup
Streamlining Python Development: A Guide to a Modern Project SetupStreamlining Python Development: A Guide to a Modern Project Setup
Streamlining Python Development: A Guide to a Modern Project SetupFlorian Wilhelm
 
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersEnhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersThousandEyes
 
Install Stable Diffusion in windows machine
Install Stable Diffusion in windows machineInstall Stable Diffusion in windows machine
Install Stable Diffusion in windows machinePadma Pradeep
 
Scanning the Internet for External Cloud Exposures via SSL Certs
Scanning the Internet for External Cloud Exposures via SSL CertsScanning the Internet for External Cloud Exposures via SSL Certs
Scanning the Internet for External Cloud Exposures via SSL CertsRizwan Syed
 
SQL Database Design For Developers at php[tek] 2024
SQL Database Design For Developers at php[tek] 2024SQL Database Design For Developers at php[tek] 2024
SQL Database Design For Developers at php[tek] 2024Scott Keck-Warren
 
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024BookNet Canada
 
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 3652toLead Limited
 
Unleash Your Potential - Namagunga Girls Coding Club
Unleash Your Potential - Namagunga Girls Coding ClubUnleash Your Potential - Namagunga Girls Coding Club
Unleash Your Potential - Namagunga Girls Coding ClubKalema Edgar
 
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...shyamraj55
 
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreterPresentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreternaman860154
 
SIEMENS: RAPUNZEL – A Tale About Knowledge Graph
SIEMENS: RAPUNZEL – A Tale About Knowledge GraphSIEMENS: RAPUNZEL – A Tale About Knowledge Graph
SIEMENS: RAPUNZEL – A Tale About Knowledge GraphNeo4j
 
Maximizing Board Effectiveness 2024 Webinar.pptx
Maximizing Board Effectiveness 2024 Webinar.pptxMaximizing Board Effectiveness 2024 Webinar.pptx
Maximizing Board Effectiveness 2024 Webinar.pptxOnBoard
 
#StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
#StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024#StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
#StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024BookNet Canada
 
Human Factors of XR: Using Human Factors to Design XR Systems
Human Factors of XR: Using Human Factors to Design XR SystemsHuman Factors of XR: Using Human Factors to Design XR Systems
Human Factors of XR: Using Human Factors to Design XR SystemsMark Billinghurst
 
Unlocking the Potential of the Cloud for IBM Power Systems
Unlocking the Potential of the Cloud for IBM Power SystemsUnlocking the Potential of the Cloud for IBM Power Systems
Unlocking the Potential of the Cloud for IBM Power SystemsPrecisely
 
08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking Men08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking MenDelhi Call girls
 

Recently uploaded (20)

Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
 
Unblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesUnblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen Frames
 
Connect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck PresentationConnect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck Presentation
 
Streamlining Python Development: A Guide to a Modern Project Setup
Streamlining Python Development: A Guide to a Modern Project SetupStreamlining Python Development: A Guide to a Modern Project Setup
Streamlining Python Development: A Guide to a Modern Project Setup
 
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersEnhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
 
Install Stable Diffusion in windows machine
Install Stable Diffusion in windows machineInstall Stable Diffusion in windows machine
Install Stable Diffusion in windows machine
 
Scanning the Internet for External Cloud Exposures via SSL Certs
Scanning the Internet for External Cloud Exposures via SSL CertsScanning the Internet for External Cloud Exposures via SSL Certs
Scanning the Internet for External Cloud Exposures via SSL Certs
 
SQL Database Design For Developers at php[tek] 2024
SQL Database Design For Developers at php[tek] 2024SQL Database Design For Developers at php[tek] 2024
SQL Database Design For Developers at php[tek] 2024
 
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
 
The transition to renewables in India.pdf
The transition to renewables in India.pdfThe transition to renewables in India.pdf
The transition to renewables in India.pdf
 
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
 
Unleash Your Potential - Namagunga Girls Coding Club
Unleash Your Potential - Namagunga Girls Coding ClubUnleash Your Potential - Namagunga Girls Coding Club
Unleash Your Potential - Namagunga Girls Coding Club
 
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
 
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreterPresentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreter
 
SIEMENS: RAPUNZEL – A Tale About Knowledge Graph
SIEMENS: RAPUNZEL – A Tale About Knowledge GraphSIEMENS: RAPUNZEL – A Tale About Knowledge Graph
SIEMENS: RAPUNZEL – A Tale About Knowledge Graph
 
Maximizing Board Effectiveness 2024 Webinar.pptx
Maximizing Board Effectiveness 2024 Webinar.pptxMaximizing Board Effectiveness 2024 Webinar.pptx
Maximizing Board Effectiveness 2024 Webinar.pptx
 
#StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
#StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024#StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
#StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
 
Human Factors of XR: Using Human Factors to Design XR Systems
Human Factors of XR: Using Human Factors to Design XR SystemsHuman Factors of XR: Using Human Factors to Design XR Systems
Human Factors of XR: Using Human Factors to Design XR Systems
 
Unlocking the Potential of the Cloud for IBM Power Systems
Unlocking the Potential of the Cloud for IBM Power SystemsUnlocking the Potential of the Cloud for IBM Power Systems
Unlocking the Potential of the Cloud for IBM Power Systems
 
08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking Men08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking Men
 

ACM ICTIR 2019 Slides - Santa Clara, USA

  • 1. Unifying Explicit and Implicit Feedback for Rating Prediction and Ranking Recommendation Tasks Amir H. Jadidinejad, Craig Macdonald, Iadh Ounis University of Glasgow ICTIR 2019
  • 2. Introduction When users visit recommendation websites they generate two types of feedback • Explicit Feedback: The user is consciously providing information to the system • Implicit Feedback: The system is collecting feedback based on users behaviour (invisibly to them) These two types of feedback can be used to automate recommendation tasks • Explicit → Rating Prediction (e.g. MF) • Implicit → Ranking (e.g. BPR) 2 Rating Reviews Likes Clicks
  • 3. Motivation 3 M. Wan and J. McAuley. Item Recommendation on Monotonic Behavior Chains, RecSys 2018. We also aim to unify explicit and implicit feedback in order to bridge the rating prediction and ranking prediction tasks click purchase review recommend feedback type
  • 4. Our Approach wrt Existing Literature Previous unifying approaches tackled the problem at the level of recommendation model • i.e. they proposed new models to combine both explicit and implicit feedback The main barrier in unifying explicit and implicit feedback is that they are heterogeneous in terms of: • Representations: <u,i,r> vs <u,i> • Distributions: 90% clicks, 10% reviews 4 Instead, we propose a novel weak supervision approach that bridges the gap between explicit and implicit feedback at the level of data preprocessing.
  • 5. Previous Unifying Approaches Models that leverage a shared representation of users and items • Joint Representation Learning for Top-N Recommendation with Heterogeneous Information Sources (Zhang et al. CIKM 2017) • Personalized Ranking Recommendation via Integrating Multiple Feedbacks (Liu et al. PAKDD 2017) Models that extend existing recommendation algorithms • Item Recommendation on Monotonic Behavior Chains. In Proceedings of (Wan and McAuley. RecSys 2018) • Bayesian Personalized Ranking with Multi-Channel User Feedback (Loni et al. RecSys 2016) 5
  • 6. Our Proposed Approach (1) 6 Our Proposal: Transfer the knowledge of rating prediction from explicit feedback to annotate implicit feedback 𝐷𝑖 = ⟨𝑈, 𝐼⟩ Idea: Can we use explicit feedback De to infer the user ratings of Di? 𝐷𝑖 ∗ = ⟨𝑈, 𝐼, 𝑅∗⟩
  • 7. Our Proposed Approach (2) 1. Train a model based on explicit feedback (ratings) : 7 Ratings Clicks
  • 8. Our Proposed Approach (2) 1. Train a model based on explicit feedback (ratings) : 2. Use the model from (1) to annotate implicit (click) feedback by predicting the rating values -> : 3. Train a new model based on the annotated dataset : 8 Ratings Predicted ratings for clicks
  • 9. Features of our Unifying Approach Approach Rating Prediction Ranking Long-tail Coverage Explicit Models (MF) Implicit Models (BPR) Our Approach 9 Our aim is to improve the capability of the explicit models for the task of ranking through the use of a novel weak supervision approach that unifies both explicit and implicit feedback datasets
  • 10. Research Questions • RQ1: Does the proposed weak supervision approach enhance recommendation accuracy for (a) Rating Prediction (b) Ranking • RQ2: How much popularity bias is exhibited by our proposed weak supervision approach compared with the other baseline recommendation approaches? 10
  • 11. Datasets 11 • GoodReads: Large book ratings dataset • Steam: Video game reviews • Douban: Chinese movie, book and music ratings • Dianping: User ratings from DianPing.com
  • 12. Evaluation Metrics RQ1: • Rating Prediction: - Root Mean Square Error (RMSE) • Ranking: - MRR, nDCG, MAP RQ2: • Long-tail item coverage: - Average Recommendation Popularity (ARP) by Yin et al. (VLDB12): 12 Reviews 80/10/10 Clicks
  • 13. Baselines 1. A popularity model 2. An explicit model - Matrix Factorization: 3. An implicit model - Bayesian Personalized Ranking: 4. An existing unifying model - ChainRec (Wan and McAuley, RecSys 2018) 13
  • 14. 14 RQ1.a Φ 𝐷 𝑒 ~ ΦDi ∗ RQ1.b ΦDi ∗ >> Φ 𝐷 𝑒 𝑀𝐹𝐵𝑃𝑅 > ΦDi ∗ RQ2 ΦDi ∗ >> 𝑀𝐹𝐵𝑃𝑅 The proposed weak supervision approach improves the capability of the explicit models for the task of ranking (5 out of 6 datasets) and has a better coverage of tail items than BPR.
  • 15. Qualitative Analysis 15 Over 70% of BPR’s recommended items are among the top-20 most popular items Only 15% of the explicit recommendations are among the top-20 most popular items (long tail coverage)
  • 16. Conclusions and Future Work Our approach unifies both the Rating prediction (explicit feedback) and Ranking (implicit feedback) tasks while alleviating the bias against less popular long-tail items. • It is applied at the level of data pre-processing as a weak supervision signal. Future Work • Combining weak supervision into pairwise approaches such as BPR or neural network-based recommenders • Tradeoff between good quality recommendations and long-tail item coverage requires more investigation. 16
  • 17. 17 Thanks! • Data and code is available at: - https://github.com/amirj/unifying_explicit_implicit