SlideShare a Scribd company logo

Interaction Design Patterns in Recommender Systems

•
•
University of Bergen
University of Bergen

Design patterns are acknowledged as powerful conceptual tools to improve design quality and to reduce the time and cost of design by effect of the reuse of “good” solutions. In many fields such as software engineering, web engineering, and interface design, patterns are widely used by practitioners and are also investigated from a research perspective. Still, the concept of design pattern has received marginal attention in the arena of user interfaces (UIs) for Recommender Systems (RSs). To our knowledge, a little is known about the use of patterns in this specific class of applications, in spite of their increasing popularity, and no RS specific interface pattern is available in existing pattern languages. We have performed a systematic analysis of 28 real-world RSs in a variety of sectors, in order to: (i) discover occurrences of existing general (i.e., domain independent) UI patterns; (ii) identify recurrent UI design solutions for RS specific features; (iii) elicit a set of new UI patterns for RS interfaces. The analysis of patterns occurrences highlights the degree at which “good” UI design solutions are adopted in RSs for the different sectors. The new patterns can be used by UI designers of RSs to improve the UX of their systems.

1 of 23
+ Interaction Design Patterns in Recommender Systems Paolo Cremonesi, Politecnico di Milano, Italy Mehdi Elahi, Politecnico di Milano, Italy Franca Garzotto, Politecnico di Milano, Italy Corresponding article: Cremonesi, Paolo; Elahi, Mehdi; Garzotto, Franca; ,Interaction Design Patterns in Recommender Systems, 11th Biannual Conference on Italian SIGCHI Chapter, 66-73, 2015, ACM
+ Outline n  Introduction: n  Recommender Systems n  Design Pattern n  Analysis n  Methodology n  Results n  Conclusion and Future Work
+ Recommender Systems tools that support users decision making by suggesting products that can be interesting to them. Examples of Recommender Systems:
+ Design Pattern: Definition “… describes a problem which occurs over and over again in our environment and then describes the core of the solution to that problem, in such a way that you can use this solution a million times over, without ever doing it the same way twice” C. Alexander. The timeless way of building, volume 1. Oxford University Press, 1979.
+ Design Pattern: Properties Main properties that the design pattern hold: n  Solves a problem: Patterns capture solutions, not just abstract principles or strategies. n  Be a proven concept: Patterns capture solutions with a track record, not theories or speculation. n  Provide a not obvious solution n  Have a significant human component
+ Design Pattern: Main Elements usually consists of the following elements: n  Problem: Problems are related to the usage of the system and are relevant to the user or any other stakeholder. n  Usage: situation(s) in which the problems occur and the pattern applies. n  Solution: a proven design solution to the problem, described in terms of design characteristics of the interface and the interaction
+ Design Pattern: Main Elements usually consists of the following elements: n  Rational: why the pattern actually works - The rationale for the solution (principles of UX quality can be used as arguments) n  Examples: how the pattern has been successfully applied in real life systems. This is often accompanied by a screenshot and a short description. n  Related Patterns: Other patterns may be needed to solve sub problems
+ Analysis: Methodology 1.  selecting 28 real-world RSs grouped by “application domain” or “business sector”; 2.  inspecting the selected RSs using a pre-defined set of user scenarios; 3.  identifying recurring design solutions; 4.  matching these solutions against existing UI patterns available in a well-established pattern library, or articulating the description of these solutions in terms of new patterns.
+ Analysis: Inspected RSs Domain RS Title Recomended Items Online Dating Meetic, Badoo, PerfectMatch User Profiles Photo Sharing Flickr, DeviantArt, Imgur, Photobucket Photos Social Bookmarking StumbleUpon, Pinterest, WeHeartIt Online content Social Network Facebook, LinkedIn, Twitter, MySpace, Google+, FourSquare User Profiles, Posts, Offers, POIs Social News Reddit, 9GAG, Digg Online content Tourism Services Booking, AirBnB, TripAdvisor, Holiday, Watchdog, Gogobot, Volagratis, Trivago, Yelp POIs R. K. Nageswara. "Application domain and functional classification of recommender systems—a survey." DESIDOC Journal of Library & Information Technology 28, no. 3 (2010): 17-35.
+ Example of Scenarios Tourism Recommender Systems: n  A young couple wants to spend their holiday in London. They would prefer to make an online reservation of their accommodation and they register to a website for online booking of hotels and bed & breakfast. They enter some information requested by the service and then receive a list of recommendations ordered by price. They choose one of them and make reservation.
+ Results: Design Pattern Usage in RSs Cremonesi, Paolo; Elahi, Mehdi; Garzotto, Franca; ,Interaction Design Patterns in Recommender Systems, 11th Biannual Conference on Italian SIGCHI Chapter, 66-73, 2015, ACM
+ Results:Why?! n  the greater the usage of design pattern the more complex is the RS n  Because the functionalities offered by Tourism Services and Online Dating systems, as well as the interaction models, are more complex in comparison to the other RSs.
+ Results: Generic Design Patterns in RSs Cremonesi, Paolo; Elahi, Mehdi; Garzotto, Franca; ,Interaction Design Patterns in Recommender Systems, 11th Biannual Conference on Italian SIGCHI Chapter, 66-73, 2015, ACM
+ Results:Why?! pagination Input prompt
+ Specific Design Patterns in RSs Social Connection Facebook Rationale: •  people are interested in doing things with their friends or with others who have similar tastes or interests •  It also provides a better estimation about what could be the users’ interests or tastes.
+ Specific Design Patterns in RSs Added Comments Deviantart Rational: •  people might be more prone to give opinion directly to other users using natural language, instead of other interaction tools, such as like/dislike or star ratings. •  The items with a large number of comments might be interesting for the community, for example, to be shown on the top of the activity stream
+ Specific Design Patterns in RSs Social Login Tripadvisor Rational: •  users may prefer to use their online profiles to avoid repeatedly entering their information other websites •  users may want to allow the system to improve the recommendation quality, by letting the system to mining their social interactions
+ Specific Design Patterns in RSs Similar Content Facebook Rationale: •  users perceive those objects that are close to each other as forming a group. •  similar content may better be shown close to the currently displayed content •  The user will likely perceive the elements as part of a group
+ Specific Design Patterns in RSs Profile as Business Card PerfectMatchbody Rationale: •  Users may enjoy f inding elements of UI that they may associate to their previous experiences (e.g. business card format) that resembles a tangible object.
+ Conclusion We have n performed a comprehensive analysis and identified a wide range of instances of existing “generic” UI design patterns n we have discovered a number of recurrent UI design problems and solutions that are specific to RSs. (No previous study has analyzed UI design patterns for RSs) n paved the ground for future research bridging RSs and Interface Design by means of design patterns which may be beneficial for RS practitioners to improve the UX quality of RSs
+ Future Work n to study the possible correlation between the design patterns adopted in RSs and the type of items recommended by them. n For example, the UI design for music RSs would be different from computer products RSs. ? Design Pattern ?
+ Future Work n to study the possible correlation between the design patterns adopted in RSs and the recommendation algorithm. n For example, the design patterns adopted in Collaborative Filtering RSs can be different from Knowledge-based RSs. ? Design Pattern ?
+ Thank you! Corresponding article: Cremonesi, Paolo; Elahi, Mehdi; Garzotto, Franca; ,Interaction Design Patterns in Recommender Systems, 11th Biannual Conference on Italian SIGCHI Chapter, 66-73, 2015, ACM

More Related Content

Interaction Design Patterns in Recommender Systems

  • 1. + Interaction Design Patterns in Recommender Systems Paolo Cremonesi, Politecnico di Milano, Italy Mehdi Elahi, Politecnico di Milano, Italy Franca Garzotto, Politecnico di Milano, Italy Corresponding article: Cremonesi, Paolo; Elahi, Mehdi; Garzotto, Franca; ,Interaction Design Patterns in Recommender Systems, 11th Biannual Conference on Italian SIGCHI Chapter, 66-73, 2015, ACM
  • 2. + Outline n  Introduction: n  Recommender Systems n  Design Pattern n  Analysis n  Methodology n  Results n  Conclusion and Future Work
  • 3. + Recommender Systems tools that support users decision making by suggesting products that can be interesting to them. Examples of Recommender Systems:
  • 4. + Design Pattern: Definition “… describes a problem which occurs over and over again in our environment and then describes the core of the solution to that problem, in such a way that you can use this solution a million times over, without ever doing it the same way twice” C. Alexander. The timeless way of building, volume 1. Oxford University Press, 1979.
  • 5. + Design Pattern: Properties Main properties that the design pattern hold: n  Solves a problem: Patterns capture solutions, not just abstract principles or strategies. n  Be a proven concept: Patterns capture solutions with a track record, not theories or speculation. n  Provide a not obvious solution n  Have a significant human component
  • 6. + Design Pattern: Main Elements usually consists of the following elements: n  Problem: Problems are related to the usage of the system and are relevant to the user or any other stakeholder. n  Usage: situation(s) in which the problems occur and the pattern applies. n  Solution: a proven design solution to the problem, described in terms of design characteristics of the interface and the interaction
  • 7. + Design Pattern: Main Elements usually consists of the following elements: n  Rational: why the pattern actually works - The rationale for the solution (principles of UX quality can be used as arguments) n  Examples: how the pattern has been successfully applied in real life systems. This is often accompanied by a screenshot and a short description. n  Related Patterns: Other patterns may be needed to solve sub problems
  • 8. + Analysis: Methodology 1.  selecting 28 real-world RSs grouped by “application domain” or “business sector”; 2.  inspecting the selected RSs using a pre-defined set of user scenarios; 3.  identifying recurring design solutions; 4.  matching these solutions against existing UI patterns available in a well-established pattern library, or articulating the description of these solutions in terms of new patterns.
  • 9. + Analysis: Inspected RSs Domain RS Title Recomended Items Online Dating Meetic, Badoo, PerfectMatch User Profiles Photo Sharing Flickr, DeviantArt, Imgur, Photobucket Photos Social Bookmarking StumbleUpon, Pinterest, WeHeartIt Online content Social Network Facebook, LinkedIn, Twitter, MySpace, Google+, FourSquare User Profiles, Posts, Offers, POIs Social News Reddit, 9GAG, Digg Online content Tourism Services Booking, AirBnB, TripAdvisor, Holiday, Watchdog, Gogobot, Volagratis, Trivago, Yelp POIs R. K. Nageswara. "Application domain and functional classification of recommender systems—a survey." DESIDOC Journal of Library & Information Technology 28, no. 3 (2010): 17-35.
  • 10. + Example of Scenarios Tourism Recommender Systems: n  A young couple wants to spend their holiday in London. They would prefer to make an online reservation of their accommodation and they register to a website for online booking of hotels and bed & breakfast. They enter some information requested by the service and then receive a list of recommendations ordered by price. They choose one of them and make reservation.
  • 11. + Results: Design Pattern Usage in RSs Cremonesi, Paolo; Elahi, Mehdi; Garzotto, Franca; ,Interaction Design Patterns in Recommender Systems, 11th Biannual Conference on Italian SIGCHI Chapter, 66-73, 2015, ACM
  • 12. + Results:Why?! n  the greater the usage of design pattern the more complex is the RS n  Because the functionalities offered by Tourism Services and Online Dating systems, as well as the interaction models, are more complex in comparison to the other RSs.
  • 13. + Results: Generic Design Patterns in RSs Cremonesi, Paolo; Elahi, Mehdi; Garzotto, Franca; ,Interaction Design Patterns in Recommender Systems, 11th Biannual Conference on Italian SIGCHI Chapter, 66-73, 2015, ACM
  • 15. + Specific Design Patterns in RSs Social Connection Facebook Rationale: •  people are interested in doing things with their friends or with others who have similar tastes or interests •  It also provides a better estimation about what could be the users’ interests or tastes.
  • 16. + Specific Design Patterns in RSs Added Comments Deviantart Rational: •  people might be more prone to give opinion directly to other users using natural language, instead of other interaction tools, such as like/dislike or star ratings. •  The items with a large number of comments might be interesting for the community, for example, to be shown on the top of the activity stream
  • 17. + Specific Design Patterns in RSs Social Login Tripadvisor Rational: •  users may prefer to use their online profiles to avoid repeatedly entering their information other websites •  users may want to allow the system to improve the recommendation quality, by letting the system to mining their social interactions
  • 18. + Specific Design Patterns in RSs Similar Content Facebook Rationale: •  users perceive those objects that are close to each other as forming a group. •  similar content may better be shown close to the currently displayed content •  The user will likely perceive the elements as part of a group
  • 19. + Specific Design Patterns in RSs Profile as Business Card PerfectMatchbody Rationale: •  Users may enjoy f inding elements of UI that they may associate to their previous experiences (e.g. business card format) that resembles a tangible object.
  • 20. + Conclusion We have n performed a comprehensive analysis and identified a wide range of instances of existing “generic” UI design patterns n we have discovered a number of recurrent UI design problems and solutions that are specific to RSs. (No previous study has analyzed UI design patterns for RSs) n paved the ground for future research bridging RSs and Interface Design by means of design patterns which may be beneficial for RS practitioners to improve the UX quality of RSs
  • 21. + Future Work n to study the possible correlation between the design patterns adopted in RSs and the type of items recommended by them. n For example, the UI design for music RSs would be different from computer products RSs. ? Design Pattern ?
  • 22. + Future Work n to study the possible correlation between the design patterns adopted in RSs and the recommendation algorithm. n For example, the design patterns adopted in Collaborative Filtering RSs can be different from Knowledge-based RSs. ? Design Pattern ?
  • 23. + Thank you! Corresponding article: Cremonesi, Paolo; Elahi, Mehdi; Garzotto, Franca; ,Interaction Design Patterns in Recommender Systems, 11th Biannual Conference on Italian SIGCHI Chapter, 66-73, 2015, ACM