Recommender Systems in E-Commerce Introduction ... - GroupLens

7 downloads 38 Views 1MB Size Report
Recommender Systems in E-Commerce. Joseph A. Konstan. John Riedl. University of Minnesota. {konstan,riedl}@cs.umn.edu http://www.cs.umn.edu/ Research/ ...
Recommender Systems in E-Commerce

ACM E-Commerce 2000

Introduction Recommender Systems in E-Commerce Joseph A. Konstan John Riedl University of Minnesota {konstan,riedl}@cs.umn.edu

What are Recommender Systems? Goals of this Tutorial Brief History of Recommender Systems

http://www.cs.umn.edu/Research/GroupLens © 2000 Joseph A. Konstan and John Riedl

ACM E-Commerce 2000

The Problem:

© 2000 Joseph A. Konstan and John Riedl

Overload

ACM E-Commerce 2000

Too much stuff! Too many books! Too many journal articles! Too many movies! Too much content!

© 2000 Joseph A. Konstan and John Riedl

ACM E-Commerce 2000

© 2000 Joseph A. Konstan and John Riedl

Scope of Recommenders

Recommenders Tools to help identify worthwhile stuff ◆ Filtering ➨ E-mail

ACM E-Commerce 2000

Purely Editorial Recommenders

interfaces filters, clipping services

◆ Recommendation ➨ Suggestion

◆ Prediction ➨ Evaluate

Content Filtering Recommenders

interfaces

lists, “top-n,” offers and promotions

interfaces

Collaborative Filtering Recommenders

candidates, predicted ratings

Hybrid Recommenders © 2000 Joseph A. Konstan and John Riedl

October 17, 2000

ACM E-Commerce 2000

2000 Joseph A. Konstan and John Riedl

ACM E-Commerce 2000

1

Recommender Systems in E-Commerce

ACM E-Commerce 2000

Goals When you leave, you should …

Tutorial Goals and Outline

2000 Joseph A. Konstan and John Riedl

ACM E-Commerce 2000

◆ Understand

recommender systems and their application to E-commerce ◆ Know enough about recommender systems technology to evaluate application ideas ◆ Be able to design and critique recommender application designs ◆ See where recommender systems have been, and where they are going 2000 Joseph A. Konstan and John Riedl

ACM E-Commerce 2000

Outline Introduction ◆

History of recommenders

History of Recommender Systems

The Virtual Shopkeeper ◆

Recommenders for online selling

MovieLens Case Study Recommender Communities The Nine Principles Application Design Model and Exercise Conclusions, Privacy, and the Future 2000 Joseph A. Konstan and John Riedl

ACM E-Commerce 2000

2000 Joseph A. Konstan and John Riedl

The Early Years … Why cave dwellers survived

ACM E-Commerce 2000

Information Filtering Information retrieval ◆ Dynamic

How editors are like cave dwellers

◆ Static

information need content base

Information filtering

Critics, critics, everywhere

◆ Static

information need content base

◆ Dynamic

2000 Joseph A. Konstan and John Riedl

October 17, 2000

ACM E-Commerce 2000

2000 Joseph A. Konstan and John Riedl

ACM E-Commerce 2000

2

Recommender Systems in E-Commerce

ACM E-Commerce 2000

Collaborative Filtering

Automated CF The GroupLens Project (CSCW ’94)

Premise ◆ Information

needs more complex than keywords or topics: quality and taste

◆ ACF

rate items are correlated with other users ➨ personal predictions for unrated items ➨ users

Small Community: Manual ◆ Tapestry

– database of content & comments ◆ Active CF – easy mechanisms for forwarding content to relevant readers

2000 Joseph A. Konstan and John Riedl

for Usenet News

➨ users

ACM E-Commerce 2000

◆ Nearest-Neighbor

Approach

➨ find

people with history of agreement ➨ assume stable tastes

2000 Joseph A. Konstan and John Riedl

ACM E-Commerce 2000

ACF Blossomed 1995 ◆ ◆

Ringo (later Firefly) Bellcore Video Recommender

1996 Recommender Systems Workshop Early commercialization Agents Inc. (later Firefly) Net Perceptions new issues of scale and performance! ◆



2000 Joseph A. Konstan and John Riedl

ACM E-Commerce 2000

2000 Joseph A. Konstan and John Riedl

Today

Introductions

Broad research community ◆ live

research systems ◆ substantial integration with machine learning, information filtering

Increasing commercial application ◆ available

commercial tools

2000 Joseph A. Konstan and John Riedl

October 17, 2000

ACM E-Commerce 2000

ACM E-Commerce 2000

John Riedl Collaborative computing, multimedia Joe Konstan User interfaces, multimedia GroupLens Research Net Perceptions 2000 Joseph A. Konstan and John Riedl

ACM E-Commerce 2000

3

Recommender Systems in E-Commerce

ACM E-Commerce 2000

The Virtual Shopkeeper

2000 Joseph A. Konstan and John Riedl

ACM E-Commerce 2000

The Consumer Is King

2000 Joseph A. Konstan and John Riedl

Carorder.com Screenshot with price information

2000 Joseph A. Konstan and John Riedl

ACM E-Commerce 2000

Book pricebot Screenshot with price information

2000 Joseph A. Konstan and John Riedl

October 17, 2000

ACM E-Commerce 2000

eBay Screenshot of mechanical clock detail page

Priceline.com Screenshot of hotel purchase with price information

2000 Joseph A. Konstan and John Riedl

ACM E-Commerce 2000

ACM E-Commerce 2000

2000 Joseph A. Konstan and John Riedl

ACM E-Commerce 2000

4