introduction and new trends in recommender systems

26
Introduction and new trends in Recommender Systems Paolo Tomeo @paotomeo

Upload: paolo-tomeo

Post on 13-Apr-2017

173 views

Category:

Data & Analytics


3 download

TRANSCRIPT

Page 1: Introduction and new trends in Recommender Systems

Introduction and new trends in Recommender Systems

Paolo Tomeo@paotomeo

Page 2: Introduction and new trends in Recommender Systems

Information overload

@mkapor

Page 3: Introduction and new trends in Recommender Systems

www.smartinsights.com/internet-marketing-statistics/happens-online-60-seconds

Page 4: Introduction and new trends in Recommender Systems

www.smartinsights.com/internet-marketing-statistics/happens-online-60-seconds

Page 5: Introduction and new trends in Recommender Systems

Recommender Systems“Software tools and techniques that provide suggestions for items that are most likely of interest to a particular user”

Page 6: Introduction and new trends in Recommender Systems
Page 7: Introduction and new trends in Recommender Systems
Page 8: Introduction and new trends in Recommender Systems
Page 9: Introduction and new trends in Recommender Systems

User – Item Interaction

Rating

Like

Click

Visualization

Search query

Page 10: Introduction and new trends in Recommender Systems

Ratings

Page 11: Introduction and new trends in Recommender Systems

Diversity matters

Page 12: Introduction and new trends in Recommender Systems

Suggerisce all'utente item simili a quelli che ha apprezzato in passato

Approaches

Content Based filtering

Collaborative filtering

Hybrid approaches

Page 13: Introduction and new trends in Recommender Systems

Suggerisce item apprezzati da altri utenti che hanno preferenze simili

Content based filtering

Recommendations based on items similar to the ones that the user liked in the past

Strengthsuser independence

explainabilityuseful for cold-start

Drawbackssensitive to bad or incomplete information

over-specializationless novelty and discovery

Page 14: Introduction and new trends in Recommender Systems

Suggerisce item apprezzati da altri utenti che hanno preferenze simili

Collaborative filtering

Recommendations based on items that other users with similar tastes liked in the past

Strengthsindependent from the content

typically more accuratecan promote discovery

Drawbackssensitive to the quantity of users and feedbacks

difficult to recommend new item (cold-start item)can reinforce item popularity

Page 15: Introduction and new trends in Recommender Systems

Suggerisce item apprezzati da altri utenti che hanno preferenze simili

Matrix factorization CF

Page 16: Introduction and new trends in Recommender Systems

Hybrid approaches

Combination of content-based and collaborative filtering methods

Ensemble of different methods Graph-based methods applied on heterogeneous networks

Feature combination -> (Matrix Factorization with side information, Factorization Machines, Neural Networks, …)

Page 17: Introduction and new trends in Recommender Systems

Beyond accuracyDiversity

Novelty

Serendipity

Explanation

Trust

Performance

Page 18: Introduction and new trends in Recommender Systems

Offline evaluation

1 - Choose a dataset2 - Split feedbacks for each user in train, validation and test sets

3 - Train the systems with the evaluation set

4 - Produce the recommendations5 - Evaluate on the test set

Page 19: Introduction and new trends in Recommender Systems

Some Libraries

RankSys - Java 8 Recommender Systems framework for novelty, diversity and much morehttps://github.com/RankSys/RankSys

Rival - Java toolkit for recommender system evaluationhttps://github.com/recommenders/rival

GraphLab Create - Python machine learning frameworkhttps://turi.com/products/create

Page 20: Introduction and new trends in Recommender Systems

(Some) New trends

Deep learning

Wide and deep learning

Multi-criteria

Graph-based algorithms

Use of Semantic Web

Page 21: Introduction and new trends in Recommender Systems

Deep learning

P. Covington, J. Adams, E. Sargin. “Deep Neural Networks for YouTube Recommendations”

Page 22: Introduction and new trends in Recommender Systems

Wide and deep learning

https://research.googleblog.com/2016/06/wide-deep-learning-better-together-with.html

https://www.tensorflow.org/versions/r0.11/tutorials/wide_and_deep/index.html

Page 23: Introduction and new trends in Recommender Systems

Multi-Criteria

Page 24: Introduction and new trends in Recommender Systems

Graph-based algorithms

Page 25: Introduction and new trends in Recommender Systems

Use of Semantic Web

Page 26: Introduction and new trends in Recommender Systems

Thanks!