recommendations as a conversation with the user

68
Recommendations as a Conversation with the User Daniel Tunkelang Principal Data Scientist at LinkedIn Danie l 1

Upload: daniel-tunkelang

Post on 09-May-2015

6.791 views

Category:

Technology


0 download

DESCRIPTION

These slides are from a tutorial at the 5th ACM International Conference on Recommender Systems (RecSys 2011). Recommender systems aim to provide users with products or content that satisfy the users' stated or inferred needs. The primary evaluation measures for recommender systems emphasize either the perceived relevance of the recommendations or the actions associated with those recommendations (e.g., purchases or clicks). Unfortunately, this transactional emphasis neglects how users interact with recommendations in the context of information seeking tasks. The effectiveness of this interaction determines the user's experience beyond a single transaction. This tutorial explores the role of recommendations as part of a conversation between the user and an information seeking system. The tutorial does not require any special background in interfaces or usability, and will focus on practical techniques to make recommender systems most effective for users.

TRANSCRIPT

Page 1: Recommendations as a Conversation with the User

1Recruiting SolutionsRecruiting SolutionsRecruiting Solutions

Recommendations as aConversation with the User

Daniel TunkelangPrincipal Data Scientist at LinkedIn

Daniel

Page 2: Recommendations as a Conversation with the User

2

Introductions

Page 3: Recommendations as a Conversation with the User

3

Let’s talk about how we talk with machines…

Page 4: Recommendations as a Conversation with the User

4

Clifford Nass’s secret:

1) Find a conclusion by a social science researcher.

2) Change

“People do X when interacting with other people.”

to

“People do X when interacting with a computer.”

3) Profit!

Page 5: Recommendations as a Conversation with the User

5

Let’s work on our relationship.

Page 6: Recommendations as a Conversation with the User

6

Core Message

Recommendations are a conversation with the user.

1) Consider asking vs. guessing.

2) Ask good questions.

3) It's ok to make mistakes…

if you have a good explanation

and adapt to feedback.

Page 7: Recommendations as a Conversation with the User

7

Our goal:

http://www.wilsoninfo.com/computerclipart.shtml

Page 8: Recommendations as a Conversation with the User

8

Overview

1) Theory

2) Examples

3) Action Items

Page 9: Recommendations as a Conversation with the User

9

1) Theory

Page 10: Recommendations as a Conversation with the User

10

Pragmatics: the Study of Conversation

Paul Grice

Page 11: Recommendations as a Conversation with the User

11

Grice’s Maxims of Conversation

Maxim 1: Quality

Maxim 2: Quantity

Maxim 3: Relation

Maxim 4: Manner

H. P. Grice, "Logic and conversation” [1975]

Page 12: Recommendations as a Conversation with the User

12

Maxim 1: Quality

Page 13: Recommendations as a Conversation with the User

13

Quality: Above All, the Truth

Xiao, Bo and Benbasat, Izak. 2011. "Product-Related Deception in E-Commerce: A Theoretical Perspective," MIS Quarterly, (35: 1) pp.169-195.

Page 14: Recommendations as a Conversation with the User

14

Don’t Lie

1) Don’t use “recommended” when you really mean

“sponsored” or “excess inventory”.

2) Optimize for the user’s utility.

3) Apply a standard of evidence (quality, quantity) that

you believe in.

Page 15: Recommendations as a Conversation with the User

15

Maxim 1: Quantity

Page 16: Recommendations as a Conversation with the User

16

Right Amount of Information

1) Exchange small units of information.

2) If recommendations supplement other content,

consider overall cognitive load.

3) Provide short, meaningful explanations.

Page 17: Recommendations as a Conversation with the User

17

Maxim 3: Relation

Page 18: Recommendations as a Conversation with the User

18

Relevant to the User

1) Offer value to the user.

2) Respect task context.

3) Don’t be obnoxious.

Page 19: Recommendations as a Conversation with the User

19

Maxim 4: Manner

Page 20: Recommendations as a Conversation with the User

20

Relevant to the User

1) Eschew obfuscation.

2) Avoid ambiguity.

3) Be brief.

4) Be orderly.

Page 21: Recommendations as a Conversation with the User

21

Another Perspective

Gary Marchionini

Page 22: Recommendations as a Conversation with the User

22

Human-Computer Information Retrieval

Empower people to explore large-scale information

but demand that

people also take responsibility for this control

by expending cognitive and physical energy.

Marchionini, G., “Toward Human-Computer Information Retrieval” [2006]

Page 23: Recommendations as a Conversation with the User

23

Principles of HCIR

1) Do more than deliver relevant information:

facilitate sensemaking.

2) Increase user responsibility and control:

require and reward effort.

3) Adapt to increasingly knowledgeable users over time.

4) Be engaging and fun to use!

Page 24: Recommendations as a Conversation with the User

24

Facilitate Sensemaking

Page 25: Recommendations as a Conversation with the User

25

Require and Reward Effort

http://www.posterenvy.com/catalog/ask_why.jpg

Page 26: Recommendations as a Conversation with the User

26

Adapt to User Knowledge

Page 27: Recommendations as a Conversation with the User

27

Be Engaging!

http://bluenile.com/

Page 28: Recommendations as a Conversation with the User

28

Applying the theory to…

1) Personalized Recommendations

2) Social Recommendations

3) Item Recommendations

Page 29: Recommendations as a Conversation with the User

29

Personalized Recommendations

1) Be transparent about model so users gain insight.

2) Allow users to modify models to correct mistakes.

3) Solicit just enough information to provide value.

Page 30: Recommendations as a Conversation with the User

30

Social Recommendations

1) Identify the right set of similar users.

2) Allow users to manipulate the social lens.

3) Accommodate users who break your model.

Page 31: Recommendations as a Conversation with the User

31

Item Recommendations

1) Explain recommendations to users.

2) Watch out for non-sequiturs (e.g., diapers -> beer).

3) Play well with user-controlled filtering and sorting.

Page 32: Recommendations as a Conversation with the User

32

2) Examples

Page 33: Recommendations as a Conversation with the User

33

Page 34: Recommendations as a Conversation with the User

34

Initial User Experience

Page 35: Recommendations as a Conversation with the User

35

“It just takes 2 minutes…”

Page 36: Recommendations as a Conversation with the User

36

Asking Before Guessing

Page 37: Recommendations as a Conversation with the User

37

Let’s try some answers:

Page 38: Recommendations as a Conversation with the User

38

Uh oh…

Page 39: Recommendations as a Conversation with the User

39

Expressing my gustibus…

Page 40: Recommendations as a Conversation with the User

40

New Star Trek = Yes; New Star Wars = No

Page 41: Recommendations as a Conversation with the User

41

Testing my patience…

Page 42: Recommendations as a Conversation with the User

42

Bring on the quality!

Page 43: Recommendations as a Conversation with the User

43

And continue the conversation.

Page 44: Recommendations as a Conversation with the User

44

Learning from Netflix

1) Ask the user for help up front. But not too much help.

2) Pay attention to what the user tells you!

3) Give users value early and often.

75% of Netflix views result from recommendations

Page 45: Recommendations as a Conversation with the User

45

Page 46: Recommendations as a Conversation with the User

46

Initial User Experience

Page 47: Recommendations as a Conversation with the User

47

Seed with an artist…

Page 48: Recommendations as a Conversation with the User

48

Or track or genre.

Page 49: Recommendations as a Conversation with the User

49

Goo Goo G'joob!

Page 50: Recommendations as a Conversation with the User

50

Ease user into recommendation space…

Page 51: Recommendations as a Conversation with the User

51

And go wild!

Page 52: Recommendations as a Conversation with the User

52

Shared Product: Personalized Stream

Page 53: Recommendations as a Conversation with the User

53

Positive and Negative Feedback

Page 54: Recommendations as a Conversation with the User

54

Learning from Pandora

1) Get meaningful input from user in one step.

2) Explain recommendations to users.

3) Solicit feedback and act on it immediately.

Page 55: Recommendations as a Conversation with the User

55

Page 56: Recommendations as a Conversation with the User

56

My home page…

Page 57: Recommendations as a Conversation with the User

57

Explanations and Humility

Page 58: Recommendations as a Conversation with the User

58

Explain What and Why

Page 59: Recommendations as a Conversation with the User

59

Recommendations as a Starting Point

Page 60: Recommendations as a Conversation with the User

60

Learning from Amazon

1) Show the factors that drive your conclusions.

2) Distinguish different kinds of recommendations.

3) Combine recommendations with user control.

Amazon: 35% of sales result from recommendations

Page 61: Recommendations as a Conversation with the User

61

3) Action Items

Page 62: Recommendations as a Conversation with the User

62

Increase explainability.

Explanations can be even more important than the recommendations themselves.

Herlocker et al., “Explaining collaborative filtering recommendations” [2000]

Sinha and Swearingen, “The role of transparency in recommender systems” [2002]

Tintarev and Masthoff, “Effective explanations of recommendations: User-centered design” [2007]

(via Òscar Celma’s book, Music Recommendation and Discovery: The Long Tail, Long Fail, and Long Play in the Digital Music Space)

Page 63: Recommendations as a Conversation with the User

63

Some models more explainable than others.

1) Consider decision trees and rule-based systems.

2) Avoid using latent, unlabeled features.

3) If the model is opaque, use examples as surrogates.

Page 64: Recommendations as a Conversation with the User

64

Make a good first impression.

Your user’s first experience is critical.

Use popularity as a default if it makes sense.

Solicit one valuable piece of information as quickly and painlessly as possible.

“Do you like the taste of beer?”

http://blog.okcupid.com/index.php/the-best-questions-for-first-dates/

Page 65: Recommendations as a Conversation with the User

65

Design feedback into your system.

You can make mistakes, if users can easily fix them.

Challenging if models use offline computation.

Respond instantly; generalize as quickly as possible.

Agarwal and Chen, “Machine Learning for Large Scale Recommender Systems” [ICML 2011 Tutorial]

Page 66: Recommendations as a Conversation with the User

66

Integrate recommendations with search.

Recommend next steps, not just items.

In a task context, recommendations are just another source of information scent.

Be careful in integrating offline recommendations with online features like search and navigation.

Pirolli, Information Foraging Theory: Adaptive Interaction with Information [2007]

Page 67: Recommendations as a Conversation with the User

67

Summary

Recommendations are a conversation with the user.

1) Consider asking vs. guessing.

2) Ask good questions.

3) It's ok to make mistakes…

if you have a good explanation

and adapt to feedback.

Page 68: Recommendations as a Conversation with the User

68

Questions?

Contact:

[email protected]

We’re Hiring!

http://engineering.linkedin.com/

Thank You!