24
Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci Morales, and Ingmar Weber Source : KDD’13 Advisor : Jia-ling Koh Speaker : Yi-hsuan Yeh

Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

Embed Size (px)

Citation preview

Page 1: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

Learning to Question: Leveraging User Preferences for

Shopping Advice

Date : 2013/12/11

Author : Mahashweta Das, Aristides Gionis,

Gianmarco De Francisci Morales,

and Ingmar Weber

Source : KDD’13

Advisor : Jia-ling Koh

Speaker : Yi-hsuan Yeh

Page 2: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

2

Outline Introduction Method Experiments Conclusion

Page 3: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

3

Introduction Motivation• Customers shop online, from their homes, without

any human interaction involved.

• Catalogs of online shops are so big and with so many continuous updates that no human, however expert, can effectively comprehend the space of available products.

Use a flowchart asks the shopper a question, and the sequence of answers leads the shopper to the suggested shopping option.

Page 4: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

4

Introduction SHOPPINGADVISOR is a novel recommender

system that helps users in shopping for technical products.

car

Page 5: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

5

Introduction SHOPPINGADVISOR generates a tree-shaped

flowchart, in which the internal nodes of the tree contain questions involve only attributes from the user space. non-expert users can understand easily.

Page 6: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

6

Introduction1. How to learn the structure of the tree, i.e., which

questions to ask at each node. Find the best user attribute to ask at each

node.

* This paper focus on identifying the attribute of interest, and not on the task of formulating the question in a human interpretable way.

2. How to produce a suitable ranking at each node. Learning-to-rank approach

Page 7: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

7

Outline Introduction Method

– LEARNSATREE algorithm Experiments Conclusion

Page 8: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

8

LEARNSATREE algorithm1. Table U (user)

2. Table P (product) 3. Table R (review)

attributes

users

Page 9: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

9

* User attributes1. Car (from Yahoo! Autos)

Ex : fuel economy, comfortable interior, stylish exterior

2. Camera (form Flickr) Photo’s tag topic

Ex : food topic (tags : fruit, market)

Page 10: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

10

Problem definition

1. Build tree 2. Rank products

node

A user attribute

Top-k list of product recommendations

Page 11: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

11

① Learning product rankings RANKSVM

Goal : Learn a weight vector for the technical attributes of the products

A > BB > C

B > D

.

.

.

RANKSVMmodel

ABDC

.

.

.

features

Product’s technical attributes

Page 12: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

12

a1 a2 a3 a4 a5

Product A 1 0 1 1 1

Product B 1 0 0 1 0

𝑤={0.2 ,0.1 ,0.5 ,0.1 ,0.1 }

rank(A)

rank(B)3

Page 13: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

13

② Learning the tree structure Goal : determine the best user attribute “” to

split at node

𝑠𝑢𝑚

Page 14: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

14

eval(rank) eval(rank)

(), (), () (), (), ()

Example : Correctly-rank :

System result

System result

Page 15: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

15

node

user

userattribut

e

split

𝑈𝑞

𝑈𝑞 (𝛼 )

𝑈𝑞 (𝛼 )

Review table

Page 16: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

16

product

RANKSVM

RANKSVM

Count payoff

ABDC...

FBEA...

Rank list

Consider all possible user attributes , and choose as splitter the one that maximizes the pay-off.

Page 17: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

17

③ Stopping criterion1) Grow the tree to its “entirety”2) Post-pruning

If a node’s child node is split by the “near-synonomous” tag trim the child node

Example: travel

vacation Employ pruning

rules on the validation set.

Page 18: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

18

Outline Introduction Method Experiments Conclusion

Page 19: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

19

Datasets1. Car datasets

• Yahoo! Autos• 606 cars, 60 attributes• 2180 reviews• 2180 user, 15 tags (as attributes)

Ex : fuel economy, comfortable interior, stylish exterior

2. Camera datasets• Flickr tags• 645 cameras (CNET)• 11468 reviews• 5647 user, 25 topic tags (as attributes)

Ex : food topic (tags : fruit, market)

3. Synthetic datasets• 200 products, 4000 comments, 1000 users

Page 20: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

20

Experiment setup1. SHOPPINGADVISOR

Author’s method

2. RANKSVM The ranked list returned by SHOPPINGADVISOR at the

root

3. k-NN k-nearest neighbors algorithm

4. SA.k-NN Features are selected from SHOPPINGADVISOR

Page 21: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

21

Quality evaluation

average MRR A

BD

.

.

.

System result ranking list

25 topics

12 topics

If user prefer “B”

Page 22: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

22

Performance evaluation

Page 23: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

23

Outline Introduction Method Experiments Conclusion

Page 24: Learning to Question: Leveraging User Preferences for Shopping Advice Date : 2013/12/11 Author : Mahashweta Das, Aristides Gionis, Gianmarco De Francisci

24

Conclusion Proposed a novel recommender system,

SHOPPINGADVISOR, that helps users to shop for technical products.

SHOPPINGADVISOR leverages both user preferences and technical product attributes in order to generate its suggestions.

At each node, SHOPPINGADVISOR suggests a ranking of products matching the preferences of the user.

Compared with a baseline, and demonstrated the effectiveness of the approach.