Designing Recommender Systems

At the heart of many products in the information age is 1 critical factor: relevance. A misguided recommendation can instantly shatter trust in your product. As a designer, how do you create a system to provide relevant information to your audience in the right place and at the right time? What is an ideal experience? In this talk, we'll look at data and design, building feedback into your system, and what you need to know about content and machine learning. We'll explore in depth case studies of designing recommendation systems and the role of "nudges" in changing behaviour and improving outcomes.

Web Directions Design 2019, Melbourne.

Link: https://didoesdigital.com/blog/talk-designing-recommender-systems-at-web-directions-design-2019/

Transcript

1. Designing Recommender Systems by Diana MacDonald // 12 Apr 2019 Designing Recommender Systems @DiDoesDigital
5. Recommender systems Designing Recommender Systems @DiDoesDigital
6. Recommender systems Predict preferences Designing Recommender Systems @DiDoesDigital
7. Recommender systems Show relevant items Designing Recommender Systems @DiDoesDigital
8. Recommender systems Content discovery and decision making Designing Recommender Systems @DiDoesDigital
9. Credit: https:/ /medium.com/netflix-techblog/artwork-personalization-c589f074ad76
12. Recommenders are handy Designing Recommender Systems @DiDoesDigital
13. The problem Designing Recommender Systems @DiDoesDigital
14. The problem Plain English Designing Recommender Systems @DiDoesDigital
15. The problem UX design Designing Recommender Systems @DiDoesDigital
16. Designing Recommender Systems @DiDoesDigital Credit: https:/ /en.wikipedia.org/wiki/Root-mean-square_deviation
 Credit: https:/ /en.wikipedia.org/wiki/Matrix_factorization_(recommender_systems)
 Credit: https:/ /commons.wikimedia.org/wiki/File:Singular_value_decomposition_visualisation.svg
18. I rarely think about looking for a job at another company Designing Recommender Systems @DiDoesDigital “ Agree
21. Designing Recommender Systems @DiDoesDigital
22. Designing recommender systems 1. Choosing outcomes 3. Improving over time 2. Design
 considerations Designing Recommender Systems @DiDoesDigital
23. Designing recommender systems 1. Choosing outcomes 3. Improving over time 2. Design
 considerations Designing Recommender Systems @DiDoesDigital
24. Choose outcomes What does success look like? Designing Recommender Systems @DiDoesDigital
25. Choose outcomes 1 of 4 / / Business impact Designing Recommender Systems @DiDoesDigital
28. Choose outcomes 2 of 4 / / Values Designing Recommender Systems @DiDoesDigital
29. How do you feel about your day at work? “
30. How do you feel about your day at work? “
31. Choose outcomes 3 of 4 / / Time to decision Designing Recommender Systems @DiDoesDigital
33. Choose outcomes 4 of 4 / / Satisfaction Designing Recommender Systems @DiDoesDigital
36. Designing recommender systems 1. Choosing outcomes 3. Improving over time 2. Design
 considerations Designing Recommender Systems @DiDoesDigital
37. Design for trust Designing Recommender Systems @DiDoesDigital
38. Design for trust Social proof Designing Recommender Systems @DiDoesDigital
40. Design for trust Labels & headings Designing Recommender Systems @DiDoesDigital
46. Design for trust Fast and slow (recency and stability) Designing Recommender Systems @DiDoesDigital
51. Design for trust Transparency Designing Recommender Systems @DiDoesDigital
54. Design for experience Designing Recommender Systems @DiDoesDigital
55. Design for experience Diversity and serendipity Designing Recommender Systems @DiDoesDigital
57. Design for experience Visual clarity Designing Recommender Systems @DiDoesDigital
58. Choosing an action
59. Choosing an action
60. Design for experience Data is not neutral Designing Recommender Systems @DiDoesDigital
64. Designing recommender systems 1. Choosing outcomes 3. Improving over time 2. Design
 considerations Designing Recommender Systems @DiDoesDigital
65. Recommender techniques Content-based filtering Collaborative filtering Designing Recommender Systems @DiDoesDigital Knowledge based
66. Recommender techniques Content-based filtering Collaborative filtering Designing Recommender Systems @DiDoesDigital Knowledge based
67. Recommender techniques Content-based filtering Collaborative filtering Designing Recommender Systems @DiDoesDigital Knowledge based
68. Recommender techniques Content-based filtering Collaborative filtering Designing Recommender Systems @DiDoesDigital Knowledge based
69. Recommender techniques Content-based filtering Collaborative filtering Designing Recommender Systems @DiDoesDigital Knowledge based
70. Recommender techniques Content-based filtering Collaborative filtering Designing Recommender Systems @DiDoesDigital Knowledge based
71. Recommender techniques Content-based filtering Collaborative filtering Designing Recommender Systems @DiDoesDigital Knowledge based
72. Recommender techniques Hybrid filtering Designing Recommender Systems @DiDoesDigital
73. Hybridization Mixed Designing Recommender Systems @DiDoesDigital
75. Feedback signals Positive and negative Designing Recommender Systems @DiDoesDigital
78. Feedback signals Implicit and explicit Designing Recommender Systems @DiDoesDigital
81. Without machine learning Designing Recommender Systems @DiDoesDigital
85. 4% started
86. 46% skipped events Games and breaks Designing Recommender Systems @DiDoesDigital
87. Designing recommender systems 1. Choosing outcomes 3. Improving over time 2. Design
 considerations Designing Recommender Systems @DiDoesDigital
88. Learn more • https:/ /uxplanet.org/how-can-we-design-an-intelligent- recommender-engine-b9bb1db4d050 • https:/ /www.nngroup.com/articles/recommendation-expectations/ • https:/ /research.netflix.com/research-area/recommendations • https:/ /en.wikipedia.org/wiki/Recommender_system Designing Recommender Systems @DiDoesDigital
89. Designing Recommender Systems @DiDoesDigital Icons by Material Design Montserrat type by Julieta Ulanovsky Questions? Diana MacDonald | DiDoesDigital.com Pre-order on Amazon