SlideShare a Scribd company logo
1 of 1
Download to read offline
1
0
0.02
0.04
0.06
0.08
P@1
DFNP
PRP
Popularity
Relevant Popularity
Average Rating
Random
Experiment
References
Recommendation optimal ranking
From the PRP to the Low Prior Discovery Recall Principle for Recommender Systems
Rocío Cañamares and Pablo Castells
Universidad Autónoma de Madrid
{Rocio.cannamares,pablo.castells}@uam.es
IRGIRGroup @UAM
 Relevant recommendations are useful, relevant and novel ones are more useful..
 New accuracy metrics
Relevant = relevant + not seen  e.g. “undiscovered precision” 𝑈𝑃@𝑘
Standard accuracy
Document search Recommendation
Query (𝑞 ∈ 𝑄) User (𝑢 ∈ 𝒰)
Document (𝑑 ∈ 𝐷) Item (𝑖 ∈ ℐ)
Relevance User taste
Relevance judgments Relevance information
Observation of taste (ratings)
DFNP: items sorted by 𝒑 𝒓𝒆𝒍 ¬𝒓𝒂𝒕𝒆𝒅, 𝒊
Undiscovered accuracy
LDRP: items sorted by 𝒑 𝒓𝒆𝒍, ¬𝒔𝒆𝒆𝒏 ¬𝒓𝒂𝒕𝒆𝒅, 𝒊
Relevance
judgments
𝐷
Relevant
Result set
PRP: documents sorted by 𝒑 𝒓𝒆𝒍 𝒅
Optimal ranking Optimal ranking Optimal ranking:
nDCG@10 Undiscovered nDCG@10
Setup
 Split all judgements into training and test (5-fold)
 Input for algorithms: training judgments on known music
 With this configuration:
 Train data → MNAR (natural biased)
 Test data → MAR (unbiased)
 Relevant = positive rating + not seen before
Dataset
Experiment
Relevance
information
Relevant
Recommended
ℐ
Ratings
𝔼 𝑃@1 = 𝑝 𝑟𝑒𝑙 𝑖1
Expected precision Expected precision
𝔼 𝑃@1 =
𝑘=1
ℐ
𝑝 𝑟𝑒𝑙, ¬𝑟𝑎𝑡𝑒𝑑 𝑖 𝑘
𝑗=1
𝑘−1
𝑝 𝑟𝑎𝑡𝑒𝑑 𝑖𝑗
Expected precision
Recommendation
Relevance
information
Relevant
Recommended
ℐ
Ratings
Seen
𝔼 𝑈𝑃@1 =
𝑘=1
ℐ
𝑝 𝑟𝑒𝑙, ¬𝑠𝑒𝑒𝑛 𝑖 𝑘
𝑗=1
𝑘−1
𝑝 𝑟𝑎𝑡𝑒𝑑 𝑖𝑗
0
20
40
60
80
100
120
Items
Undiscovered accuracy
We build a dataset with unbiased observations (judgments)
 We sample 1,000 music tracks from deezer.com uniformly at random
 We ask ~1,000 anonymous workers on CrowdFlower to rate 100 tracks
each, sampled uniformly at random → ~100,000 ratings total
Available at http://ir.ii.uam.es/cm100k
Setup
 Training data: random subsample (ratio 𝜌) of judgments
on known music → MNAR (naturally biased)
 Relevance information: all remaining judgments
Typical offline evaluation setup
 All available rating data is split into training and test.
 Training data are given to recommenders → Ratings
 Test data are used for evaluating the recommendation → Relevance information
Problem
 The relevance information is biased: users rate what they want, typically some
items have more ratings than others
Goal
 Collect unbiased relevance information for evaluation
Judgments
𝜌
0
1
Relevance
informationRatings
Items
Unknown
music
Known
music
Relevance
informationRatings
Known music
Unknown music
Known music
𝜌 = 0.8
Items
0
0.02
0.04
0.06
0.1 1
nDCG@10
r
PRP and popularities very
close to optimal ranking
Average rating seems
to underperform
considerably PRP and popularities
worse than random!
The comparison between algorithms changes completely when we consider relevant and novel recommendations
0
0.01
0.02
0
0.01
0.02
Results
 DFNP more robust than PRP to the split ratio
⁻ PRP worse than random when 𝜌 = 1
 Popularity seems to follow PRP
 Average rating better than popularity when 𝜌 is close to 1
 Potential implications on state of the art personalized algorithms
[1] A. Bellogín, P. Castells, and I. Cantador. Statistical Biases in Information Retrieval Metrics for Recommender Systems. Inf. Ret. 20, 6 (Dec. 2017), 606–634
[2] R. Cañamares and P. Castells. 2017. A Probabilistic Reformulation of Memory-Based Collaborative Filtering – Implications on Popularity Biases. In Proceedings of the 40th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2017), 215–224
[3] P. Castells, N. J. Hurley, S. Vargas. 2015. Novelty and Diversity in Recommender Systems. In: Recommender Systems Handbook, 2nd edition, F. Ricci, L. Rokach, and B. Shapira (Eds.). Springer, New York, NY, USA, 881–918
[4] P. Cremonesi, Y. Koren, and R. Turrin. 2010. Performance of recommender algorithms on top-n recommendation tasks. In Proceedings of the 4th ACM Conference on Recommender Systems (RecSys 2010), 39–46
[5] S. E. Robertson. 1977. The Probability Ranking in IR. Journal of Documentation 33, 4 (Jan. 1977), 294–304.
[6] H. Steck. 2011. Item popularity and recommendation accuracy. In Proc. of the 5th ACM Conference on Recommender Systems (RecSys 2011), 125–132
[7] H. Steck. 2010. Training and testing of recommender systems on data missing not at random. In Proc. of the 16th ACM SIGKDD Int. Conf. on Knowledge Discovery and Data Mining (KDD 2010). ACM, New York, NY, USA, 713–722
[8] B. Marlin and R. Zemel. 2009. Collaborative prediction and ranking with non-random missing data. In Proc. of the 3rd ACM Conference on Recommender Systems (RecSys 2009). ACM, New York, NY, USA, 5–12
41st Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2018)
Average rating is
the only algorithm
standing above random

More Related Content

What's hot (6)

Upgrading the Scholarly Infrastructure
Upgrading the Scholarly InfrastructureUpgrading the Scholarly Infrastructure
Upgrading the Scholarly Infrastructure
 
How to do research juvenile delinquency 2013
How to do research juvenile delinquency 2013How to do research juvenile delinquency 2013
How to do research juvenile delinquency 2013
 
Whether in a scholarly or practitioner setting, good research and da
Whether in a scholarly or practitioner setting, good research and daWhether in a scholarly or practitioner setting, good research and da
Whether in a scholarly or practitioner setting, good research and da
 
COMM1180-Alcock November 2012
COMM1180-Alcock November 2012COMM1180-Alcock November 2012
COMM1180-Alcock November 2012
 
The Human Factor in Digital Recommender Systems
The Human Factor in Digital Recommender SystemsThe Human Factor in Digital Recommender Systems
The Human Factor in Digital Recommender Systems
 
Surveys SD
Surveys SDSurveys SD
Surveys SD
 

Similar to SIGIR 2018 - From the Probability Ranking Principle to the Low Prior Discovery Recall Principle for Recommender Systems

Due Wedneday 900 pm New York TimeData Analysis for Qualitative In.docx
Due Wedneday 900 pm New York TimeData Analysis for Qualitative In.docxDue Wedneday 900 pm New York TimeData Analysis for Qualitative In.docx
Due Wedneday 900 pm New York TimeData Analysis for Qualitative In.docx
shandicollingwood
 
HRCU 351 Week 6 Paper 2 EthicsGuidelines and Grading Rubric
HRCU 351 Week 6 Paper 2 EthicsGuidelines and Grading RubricHRCU 351 Week 6 Paper 2 EthicsGuidelines and Grading Rubric
HRCU 351 Week 6 Paper 2 EthicsGuidelines and Grading Rubric
PazSilviapm
 
DESCRIPTIVE ANALYSIS1DESCRIPTIVE ANALYSIS8Examining .docx
DESCRIPTIVE ANALYSIS1DESCRIPTIVE ANALYSIS8Examining .docxDESCRIPTIVE ANALYSIS1DESCRIPTIVE ANALYSIS8Examining .docx
DESCRIPTIVE ANALYSIS1DESCRIPTIVE ANALYSIS8Examining .docx
donaldp2
 

Similar to SIGIR 2018 - From the Probability Ranking Principle to the Low Prior Discovery Recall Principle for Recommender Systems (20)

Information Retrieval Models for Recommender Systems - PhD slides
Information Retrieval Models for Recommender Systems - PhD slidesInformation Retrieval Models for Recommender Systems - PhD slides
Information Retrieval Models for Recommender Systems - PhD slides
 
Haystack keynote 2019: What is Search Relevance? - Max Irwin
Haystack keynote 2019: What is Search Relevance? - Max IrwinHaystack keynote 2019: What is Search Relevance? - Max Irwin
Haystack keynote 2019: What is Search Relevance? - Max Irwin
 
uyyu.pptx
uyyu.pptxuyyu.pptx
uyyu.pptx
 
AI Driven Product Innovation
AI Driven Product InnovationAI Driven Product Innovation
AI Driven Product Innovation
 
AI-driven product innovation: from Recommender Systems to COVID-19
AI-driven product innovation: from Recommender Systems to COVID-19AI-driven product innovation: from Recommender Systems to COVID-19
AI-driven product innovation: from Recommender Systems to COVID-19
 
Due Wedneday 900 pm New York TimeData Analysis for Qualitative In.docx
Due Wedneday 900 pm New York TimeData Analysis for Qualitative In.docxDue Wedneday 900 pm New York TimeData Analysis for Qualitative In.docx
Due Wedneday 900 pm New York TimeData Analysis for Qualitative In.docx
 
An introduction to system-oriented evaluation in Information Retrieval
An introduction to system-oriented evaluation in Information RetrievalAn introduction to system-oriented evaluation in Information Retrieval
An introduction to system-oriented evaluation in Information Retrieval
 
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...
 
HRCU 351 Week 6 Paper 2 EthicsGuidelines and Grading Rubric
HRCU 351 Week 6 Paper 2 EthicsGuidelines and Grading RubricHRCU 351 Week 6 Paper 2 EthicsGuidelines and Grading Rubric
HRCU 351 Week 6 Paper 2 EthicsGuidelines and Grading Rubric
 
Research Data Management
Research  Data ManagementResearch  Data Management
Research Data Management
 
sigir2020
sigir2020sigir2020
sigir2020
 
April Heyward Research Methods Class Session - 7-29-2021
April Heyward Research Methods Class Session - 7-29-2021April Heyward Research Methods Class Session - 7-29-2021
April Heyward Research Methods Class Session - 7-29-2021
 
Social Recommender Systems
Social Recommender SystemsSocial Recommender Systems
Social Recommender Systems
 
The International Journal of Engineering and Science (The IJES)
The International Journal of Engineering and Science (The IJES)The International Journal of Engineering and Science (The IJES)
The International Journal of Engineering and Science (The IJES)
 
Tutorial: Context-awareness In Information Retrieval and Recommender Systems
Tutorial: Context-awareness In Information Retrieval and Recommender SystemsTutorial: Context-awareness In Information Retrieval and Recommender Systems
Tutorial: Context-awareness In Information Retrieval and Recommender Systems
 
Search Quality Evaluation to Help Reproducibility: An Open-source Approach
Search Quality Evaluation to Help Reproducibility: An Open-source ApproachSearch Quality Evaluation to Help Reproducibility: An Open-source Approach
Search Quality Evaluation to Help Reproducibility: An Open-source Approach
 
Researcher perspectives on publication and peer review of data.
Researcher perspectives on publication and peer review of data.Researcher perspectives on publication and peer review of data.
Researcher perspectives on publication and peer review of data.
 
Search Quality Evaluation to Help Reproducibility : an Open Source Approach
Search Quality Evaluation to Help Reproducibility : an Open Source ApproachSearch Quality Evaluation to Help Reproducibility : an Open Source Approach
Search Quality Evaluation to Help Reproducibility : an Open Source Approach
 
quantitative vs qualitative
quantitative vs qualitativequantitative vs qualitative
quantitative vs qualitative
 
DESCRIPTIVE ANALYSIS1DESCRIPTIVE ANALYSIS8Examining .docx
DESCRIPTIVE ANALYSIS1DESCRIPTIVE ANALYSIS8Examining .docxDESCRIPTIVE ANALYSIS1DESCRIPTIVE ANALYSIS8Examining .docx
DESCRIPTIVE ANALYSIS1DESCRIPTIVE ANALYSIS8Examining .docx
 

Recently uploaded

Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service BangaloreCall Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
amitlee9823
 
Call Girls In Doddaballapur Road ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Doddaballapur Road ☎ 7737669865 🥵 Book Your One night StandCall Girls In Doddaballapur Road ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Doddaballapur Road ☎ 7737669865 🥵 Book Your One night Stand
amitlee9823
 
Just Call Vip call girls roorkee Escorts ☎️9352988975 Two shot with one girl ...
Just Call Vip call girls roorkee Escorts ☎️9352988975 Two shot with one girl ...Just Call Vip call girls roorkee Escorts ☎️9352988975 Two shot with one girl ...
Just Call Vip call girls roorkee Escorts ☎️9352988975 Two shot with one girl ...
gajnagarg
 
👉 Amritsar Call Girl 👉📞 6367187148 👉📞 Just📲 Call Ruhi Call Girl Phone No Amri...
👉 Amritsar Call Girl 👉📞 6367187148 👉📞 Just📲 Call Ruhi Call Girl Phone No Amri...👉 Amritsar Call Girl 👉📞 6367187148 👉📞 Just📲 Call Ruhi Call Girl Phone No Amri...
👉 Amritsar Call Girl 👉📞 6367187148 👉📞 Just📲 Call Ruhi Call Girl Phone No Amri...
karishmasinghjnh
 
Abortion pills in Jeddah | +966572737505 | Get Cytotec
Abortion pills in Jeddah | +966572737505 | Get CytotecAbortion pills in Jeddah | +966572737505 | Get Cytotec
Abortion pills in Jeddah | +966572737505 | Get Cytotec
Abortion pills in Riyadh +966572737505 get cytotec
 
➥🔝 7737669865 🔝▻ Bangalore Call-girls in Women Seeking Men 🔝Bangalore🔝 Esc...
➥🔝 7737669865 🔝▻ Bangalore Call-girls in Women Seeking Men  🔝Bangalore🔝   Esc...➥🔝 7737669865 🔝▻ Bangalore Call-girls in Women Seeking Men  🔝Bangalore🔝   Esc...
➥🔝 7737669865 🔝▻ Bangalore Call-girls in Women Seeking Men 🔝Bangalore🔝 Esc...
amitlee9823
 
Chintamani Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore ...
Chintamani Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore ...Chintamani Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore ...
Chintamani Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore ...
amitlee9823
 
Call Girls Bommasandra Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Bommasandra Just Call 👗 7737669865 👗 Top Class Call Girl Service B...Call Girls Bommasandra Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Bommasandra Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
amitlee9823
 
Vip Mumbai Call Girls Thane West Call On 9920725232 With Body to body massage...
Vip Mumbai Call Girls Thane West Call On 9920725232 With Body to body massage...Vip Mumbai Call Girls Thane West Call On 9920725232 With Body to body massage...
Vip Mumbai Call Girls Thane West Call On 9920725232 With Body to body massage...
amitlee9823
 
Abortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Doha Qatar (+966572737505 ! Get CytotecAbortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Riyadh +966572737505 get cytotec
 
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...
amitlee9823
 
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
amitlee9823
 
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
amitlee9823
 
Just Call Vip call girls Palakkad Escorts ☎️9352988975 Two shot with one girl...
Just Call Vip call girls Palakkad Escorts ☎️9352988975 Two shot with one girl...Just Call Vip call girls Palakkad Escorts ☎️9352988975 Two shot with one girl...
Just Call Vip call girls Palakkad Escorts ☎️9352988975 Two shot with one girl...
gajnagarg
 
Vip Mumbai Call Girls Marol Naka Call On 9920725232 With Body to body massage...
Vip Mumbai Call Girls Marol Naka Call On 9920725232 With Body to body massage...Vip Mumbai Call Girls Marol Naka Call On 9920725232 With Body to body massage...
Vip Mumbai Call Girls Marol Naka Call On 9920725232 With Body to body massage...
amitlee9823
 
Call Girls In Nandini Layout ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Nandini Layout ☎ 7737669865 🥵 Book Your One night StandCall Girls In Nandini Layout ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Nandini Layout ☎ 7737669865 🥵 Book Your One night Stand
amitlee9823
 
Call Girls In Bellandur ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Bellandur ☎ 7737669865 🥵 Book Your One night StandCall Girls In Bellandur ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Bellandur ☎ 7737669865 🥵 Book Your One night Stand
amitlee9823
 
➥🔝 7737669865 🔝▻ Ongole Call-girls in Women Seeking Men 🔝Ongole🔝 Escorts S...
➥🔝 7737669865 🔝▻ Ongole Call-girls in Women Seeking Men  🔝Ongole🔝   Escorts S...➥🔝 7737669865 🔝▻ Ongole Call-girls in Women Seeking Men  🔝Ongole🔝   Escorts S...
➥🔝 7737669865 🔝▻ Ongole Call-girls in Women Seeking Men 🔝Ongole🔝 Escorts S...
amitlee9823
 

Recently uploaded (20)

Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service BangaloreCall Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
 
VIP Model Call Girls Hinjewadi ( Pune ) Call ON 8005736733 Starting From 5K t...
VIP Model Call Girls Hinjewadi ( Pune ) Call ON 8005736733 Starting From 5K t...VIP Model Call Girls Hinjewadi ( Pune ) Call ON 8005736733 Starting From 5K t...
VIP Model Call Girls Hinjewadi ( Pune ) Call ON 8005736733 Starting From 5K t...
 
Call Girls In Doddaballapur Road ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Doddaballapur Road ☎ 7737669865 🥵 Book Your One night StandCall Girls In Doddaballapur Road ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Doddaballapur Road ☎ 7737669865 🥵 Book Your One night Stand
 
Just Call Vip call girls roorkee Escorts ☎️9352988975 Two shot with one girl ...
Just Call Vip call girls roorkee Escorts ☎️9352988975 Two shot with one girl ...Just Call Vip call girls roorkee Escorts ☎️9352988975 Two shot with one girl ...
Just Call Vip call girls roorkee Escorts ☎️9352988975 Two shot with one girl ...
 
👉 Amritsar Call Girl 👉📞 6367187148 👉📞 Just📲 Call Ruhi Call Girl Phone No Amri...
👉 Amritsar Call Girl 👉📞 6367187148 👉📞 Just📲 Call Ruhi Call Girl Phone No Amri...👉 Amritsar Call Girl 👉📞 6367187148 👉📞 Just📲 Call Ruhi Call Girl Phone No Amri...
👉 Amritsar Call Girl 👉📞 6367187148 👉📞 Just📲 Call Ruhi Call Girl Phone No Amri...
 
Abortion pills in Jeddah | +966572737505 | Get Cytotec
Abortion pills in Jeddah | +966572737505 | Get CytotecAbortion pills in Jeddah | +966572737505 | Get Cytotec
Abortion pills in Jeddah | +966572737505 | Get Cytotec
 
➥🔝 7737669865 🔝▻ Bangalore Call-girls in Women Seeking Men 🔝Bangalore🔝 Esc...
➥🔝 7737669865 🔝▻ Bangalore Call-girls in Women Seeking Men  🔝Bangalore🔝   Esc...➥🔝 7737669865 🔝▻ Bangalore Call-girls in Women Seeking Men  🔝Bangalore🔝   Esc...
➥🔝 7737669865 🔝▻ Bangalore Call-girls in Women Seeking Men 🔝Bangalore🔝 Esc...
 
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...
 
Chintamani Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore ...
Chintamani Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore ...Chintamani Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore ...
Chintamani Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore ...
 
Call Girls Bommasandra Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Bommasandra Just Call 👗 7737669865 👗 Top Class Call Girl Service B...Call Girls Bommasandra Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Bommasandra Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
 
Vip Mumbai Call Girls Thane West Call On 9920725232 With Body to body massage...
Vip Mumbai Call Girls Thane West Call On 9920725232 With Body to body massage...Vip Mumbai Call Girls Thane West Call On 9920725232 With Body to body massage...
Vip Mumbai Call Girls Thane West Call On 9920725232 With Body to body massage...
 
Abortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Doha Qatar (+966572737505 ! Get CytotecAbortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Doha Qatar (+966572737505 ! Get Cytotec
 
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...
 
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
 
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
 
Just Call Vip call girls Palakkad Escorts ☎️9352988975 Two shot with one girl...
Just Call Vip call girls Palakkad Escorts ☎️9352988975 Two shot with one girl...Just Call Vip call girls Palakkad Escorts ☎️9352988975 Two shot with one girl...
Just Call Vip call girls Palakkad Escorts ☎️9352988975 Two shot with one girl...
 
Vip Mumbai Call Girls Marol Naka Call On 9920725232 With Body to body massage...
Vip Mumbai Call Girls Marol Naka Call On 9920725232 With Body to body massage...Vip Mumbai Call Girls Marol Naka Call On 9920725232 With Body to body massage...
Vip Mumbai Call Girls Marol Naka Call On 9920725232 With Body to body massage...
 
Call Girls In Nandini Layout ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Nandini Layout ☎ 7737669865 🥵 Book Your One night StandCall Girls In Nandini Layout ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Nandini Layout ☎ 7737669865 🥵 Book Your One night Stand
 
Call Girls In Bellandur ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Bellandur ☎ 7737669865 🥵 Book Your One night StandCall Girls In Bellandur ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Bellandur ☎ 7737669865 🥵 Book Your One night Stand
 
➥🔝 7737669865 🔝▻ Ongole Call-girls in Women Seeking Men 🔝Ongole🔝 Escorts S...
➥🔝 7737669865 🔝▻ Ongole Call-girls in Women Seeking Men  🔝Ongole🔝   Escorts S...➥🔝 7737669865 🔝▻ Ongole Call-girls in Women Seeking Men  🔝Ongole🔝   Escorts S...
➥🔝 7737669865 🔝▻ Ongole Call-girls in Women Seeking Men 🔝Ongole🔝 Escorts S...
 

SIGIR 2018 - From the Probability Ranking Principle to the Low Prior Discovery Recall Principle for Recommender Systems

  • 1. 1 0 0.02 0.04 0.06 0.08 P@1 DFNP PRP Popularity Relevant Popularity Average Rating Random Experiment References Recommendation optimal ranking From the PRP to the Low Prior Discovery Recall Principle for Recommender Systems Rocío Cañamares and Pablo Castells Universidad Autónoma de Madrid {Rocio.cannamares,pablo.castells}@uam.es IRGIRGroup @UAM  Relevant recommendations are useful, relevant and novel ones are more useful..  New accuracy metrics Relevant = relevant + not seen  e.g. “undiscovered precision” 𝑈𝑃@𝑘 Standard accuracy Document search Recommendation Query (𝑞 ∈ 𝑄) User (𝑢 ∈ 𝒰) Document (𝑑 ∈ 𝐷) Item (𝑖 ∈ ℐ) Relevance User taste Relevance judgments Relevance information Observation of taste (ratings) DFNP: items sorted by 𝒑 𝒓𝒆𝒍 ¬𝒓𝒂𝒕𝒆𝒅, 𝒊 Undiscovered accuracy LDRP: items sorted by 𝒑 𝒓𝒆𝒍, ¬𝒔𝒆𝒆𝒏 ¬𝒓𝒂𝒕𝒆𝒅, 𝒊 Relevance judgments 𝐷 Relevant Result set PRP: documents sorted by 𝒑 𝒓𝒆𝒍 𝒅 Optimal ranking Optimal ranking Optimal ranking: nDCG@10 Undiscovered nDCG@10 Setup  Split all judgements into training and test (5-fold)  Input for algorithms: training judgments on known music  With this configuration:  Train data → MNAR (natural biased)  Test data → MAR (unbiased)  Relevant = positive rating + not seen before Dataset Experiment Relevance information Relevant Recommended ℐ Ratings 𝔼 𝑃@1 = 𝑝 𝑟𝑒𝑙 𝑖1 Expected precision Expected precision 𝔼 𝑃@1 = 𝑘=1 ℐ 𝑝 𝑟𝑒𝑙, ¬𝑟𝑎𝑡𝑒𝑑 𝑖 𝑘 𝑗=1 𝑘−1 𝑝 𝑟𝑎𝑡𝑒𝑑 𝑖𝑗 Expected precision Recommendation Relevance information Relevant Recommended ℐ Ratings Seen 𝔼 𝑈𝑃@1 = 𝑘=1 ℐ 𝑝 𝑟𝑒𝑙, ¬𝑠𝑒𝑒𝑛 𝑖 𝑘 𝑗=1 𝑘−1 𝑝 𝑟𝑎𝑡𝑒𝑑 𝑖𝑗 0 20 40 60 80 100 120 Items Undiscovered accuracy We build a dataset with unbiased observations (judgments)  We sample 1,000 music tracks from deezer.com uniformly at random  We ask ~1,000 anonymous workers on CrowdFlower to rate 100 tracks each, sampled uniformly at random → ~100,000 ratings total Available at http://ir.ii.uam.es/cm100k Setup  Training data: random subsample (ratio 𝜌) of judgments on known music → MNAR (naturally biased)  Relevance information: all remaining judgments Typical offline evaluation setup  All available rating data is split into training and test.  Training data are given to recommenders → Ratings  Test data are used for evaluating the recommendation → Relevance information Problem  The relevance information is biased: users rate what they want, typically some items have more ratings than others Goal  Collect unbiased relevance information for evaluation Judgments 𝜌 0 1 Relevance informationRatings Items Unknown music Known music Relevance informationRatings Known music Unknown music Known music 𝜌 = 0.8 Items 0 0.02 0.04 0.06 0.1 1 nDCG@10 r PRP and popularities very close to optimal ranking Average rating seems to underperform considerably PRP and popularities worse than random! The comparison between algorithms changes completely when we consider relevant and novel recommendations 0 0.01 0.02 0 0.01 0.02 Results  DFNP more robust than PRP to the split ratio ⁻ PRP worse than random when 𝜌 = 1  Popularity seems to follow PRP  Average rating better than popularity when 𝜌 is close to 1  Potential implications on state of the art personalized algorithms [1] A. Bellogín, P. Castells, and I. Cantador. Statistical Biases in Information Retrieval Metrics for Recommender Systems. Inf. Ret. 20, 6 (Dec. 2017), 606–634 [2] R. Cañamares and P. Castells. 2017. A Probabilistic Reformulation of Memory-Based Collaborative Filtering – Implications on Popularity Biases. In Proceedings of the 40th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2017), 215–224 [3] P. Castells, N. J. Hurley, S. Vargas. 2015. Novelty and Diversity in Recommender Systems. In: Recommender Systems Handbook, 2nd edition, F. Ricci, L. Rokach, and B. Shapira (Eds.). Springer, New York, NY, USA, 881–918 [4] P. Cremonesi, Y. Koren, and R. Turrin. 2010. Performance of recommender algorithms on top-n recommendation tasks. In Proceedings of the 4th ACM Conference on Recommender Systems (RecSys 2010), 39–46 [5] S. E. Robertson. 1977. The Probability Ranking in IR. Journal of Documentation 33, 4 (Jan. 1977), 294–304. [6] H. Steck. 2011. Item popularity and recommendation accuracy. In Proc. of the 5th ACM Conference on Recommender Systems (RecSys 2011), 125–132 [7] H. Steck. 2010. Training and testing of recommender systems on data missing not at random. In Proc. of the 16th ACM SIGKDD Int. Conf. on Knowledge Discovery and Data Mining (KDD 2010). ACM, New York, NY, USA, 713–722 [8] B. Marlin and R. Zemel. 2009. Collaborative prediction and ranking with non-random missing data. In Proc. of the 3rd ACM Conference on Recommender Systems (RecSys 2009). ACM, New York, NY, USA, 5–12 41st Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2018) Average rating is the only algorithm standing above random