how to build a recommender system

38
Recommender System How to build a

Upload: vo-duy-tuan

Post on 18-May-2015

4.108 views

Category:

Technology


0 download

DESCRIPTION

This presentation show the method to build a Recommender System with Collaborative FIltering method.

TRANSCRIPT

Page 1: How to build a Recommender System

Recommender System

How to build a

Page 2: How to build a Recommender System

Võ Duy TuấnTechnical Director @ dienmay.com

PHP 5 Zend Certified Engineer

Mobile App Developer

Web Developer & Designer

Interest: o PHP

o Large System & Data Mining

o Web Performance Optimization

o Mobile Development

Page 3: How to build a Recommender System

Introduction

Collaborative Filtering

Question & Answer

AGENDA

Page 4: How to build a Recommender System

1. Introduction

Page 5: How to build a Recommender System

APPLICATIONS

• Personalized recommendation• Social recommendation• Item recommendation• Combination of 3 approaches above

Page 6: How to build a Recommender System

AMAZON.COM | BOOKS

Page 7: How to build a Recommender System

PLAY.GOOGLE.COM | APPS

Page 8: How to build a Recommender System

SKILLSHARE.COM | CLASSES

Page 9: How to build a Recommender System

PROCESS DIAGRAM

Preprocessing Data Analysis Adjustment

INPUT OUTPUT

Page 10: How to build a Recommender System

TYPE OF RECOMMENDER SYSTEM

• Collaborative filtering• Content-based filtering• Hybrid

Page 11: How to build a Recommender System

2. Collaborative Filtering

Page 12: How to build a Recommender System

USER & ITEM

Page 13: How to build a Recommender System

ORDER DATA

Page 14: How to build a Recommender System

ORDER DATA (cont.)

Page 15: How to build a Recommender System

ORDER DATA (cont.)

Page 16: How to build a Recommender System

VECTOR & DIMENSION

Page 17: How to build a Recommender System

VECTOR & DIMENSION

Page 18: How to build a Recommender System

VECTORS

Page 19: How to build a Recommender System

VECTORS

Page 20: How to build a Recommender System

SIMILARITY CALCULATION

Page 21: How to build a Recommender System

USER SIMILARITY MATRIX

Page 22: How to build a Recommender System

SIMILARITY CALCULATION

Page 23: How to build a Recommender System

SIMILARITY CALCULATION

Page 24: How to build a Recommender System

SIMILARITY CALCULATION EXAMPLE

Page 25: How to build a Recommender System

K-NEAREST-NEIGHBOR

Page 26: How to build a Recommender System

K-NEAREST-NEIGHBOR

Page 27: How to build a Recommender System

NEIGHBORS’ ORDER

Page 28: How to build a Recommender System

REMOVE BOUGHT ITEMS

Page 29: How to build a Recommender System

CALCULATING FINAL SCORE

Page 30: How to build a Recommender System

OTHER SIMILARITY MEASURES

More at: http://favi.com.vn/wp-content/uploads/2012/05/pg049_Similarity_Measures_for_Text_Document_Clustering.pdf

Page 31: How to build a Recommender System

Problem ?!

Page 32: How to build a Recommender System

COLLABORATIVE FILTERING PROBLEM

• Fail with cold start problemo New Usero New Item

• Performanceo Large Data seto Pre-calculate

Page 33: How to build a Recommender System

PERFORMANCE EXAMPLE

• We have 1,000,000 users (customers)• We sell 10,000 items

- Total of similarity calculating = 1,000,000 x 1,000,000 = 1,000,000,000,000- Each similarity calculate need 0.006s (on my MacBook Pro 2.2GHz Core i7, 8G Ram)

=> We need 1,000,000,000,000 x 0.006 = 6,000,000,000(s)≈ 70,000 days ≈ 191 years

- If store each similarity in 8 bytes, we need = 8,000,000,000,000 bytes≈ 8,000 GB (on Memory or File)

Page 34: How to build a Recommender System

ITEM-TO-ITEM COLLABORATIVE FILTERING (AMAZON.COM )

Download Paper: http://www.cs.umd.edu/~samir/498/Amazon-Recommendations.pdf

Page 35: How to build a Recommender System

ADJUSTMENTS

• Hybrid Recommender System• Sale forecast system• Context of User• Type of Item, Action• External (3rd-party) information.

Page 36: How to build a Recommender System

BOOKS

Programming Collective IntelligenceToby Segaran

Recommender Systems HandbookMany Authors

Big Data For DummiesMarcia Kaufman, Fern Halper

Page 37: How to build a Recommender System

OPEN SOURCES

Page 38: How to build a Recommender System

Thank you!

CONTACT ME:

[email protected]

0938 916 902

http://bloghoctap.com/