Tools for decision making and problem solving: Choice (e.g., Comparison shopping, recommender systems)
Download ReportTranscript Tools for decision making and problem solving: Choice (e.g., Comparison shopping, recommender systems)
Am Introduction to Recommendation Systems Hasan Davulcu CIPS, ASU Recommendation Systems U: C X S R C: profile: age, gender, income S: title, director, actor, year, genre Recommendations: Recommender Systems Content Based Limitations Too Similar ! New user problem User - Collaborative Methods U(C,S) is estimated based on utilities U(Cj,S) by those users Cj who are “similar” to user C. Limitations New Item Problem Sparsity Amazon’s Item-to-Item Comparing Human Recommenders to Online Systems Rashmi Sinha & Kirsten Swearingen SIMS, UC Berkeley Recommender Systems are technological proxy for a social process Which one should I read? Recommendations from friends Recommendations from Online Systems I know what you’ll read next summer (Amazon, Barnes&Noble) what movies you should watch… (Reel, RatingZone, Amazon) what music you should listen to… (CDNow, Mubu, Gigabeat) what websites you should visit (Alexa) what jokes you will like (Jester) & who you should date (Yenta) Method Philosophy: Testing & Analysis as part of the Iterative Design Process Design Evaluate Use both quantitative & qualitative methods Analyze Slide adapted from James Landay Generate Design Recommendations Taking a closer look at the Recommendation Process Input User incurs cost in using system: Time, Effort, Privacy Issues Receives Recommendation Cost in reviewing recommendations Judges if he/she will sample recommendation Benefit only if recommended item appeals Amazon’s Recommendation Process Input: One artist/author name Search using Recommendations Output: List of Recommendations Explore / Refine Recommendations Book Recommendation Site: Sleeper Input: Ratings of 10 books for all users Use of continuous Rating Bar (System designed by Ken Goldberg) Sleeper: Output Output: List of items with brief information about each item Degree of confidence in prediction What convinces a user to sample the recommendation Judging recommendations: Trust in a Recommender System: What is a good recommendation from the user’s perspective? What factors lead to trust in a system? System Transparency: Do users need to know why an item was recommended? Study of RS has focused mostly on Collaborative Filtering Algorithms Social Recommendations Input from user Output (Recommendations) Collaborative Filtering Algorithms Beyond “Algorithms Only” : An HCI Perspective on Recommender Systems Comparing the Social Recommendation Process to Online Recommender Systems Understanding the factors that go into an effective recommendation (by studying users interaction with 6 online RS) The Human vs. Recommenders Death Match Book Systems Amazon Books Rating Zone Sleeper Movie Systems Amazon Movies Movie Critic Reel Method 19 For each of 3 online systems: participants, age:18 to 34 years Registered at site Rated items Reviewed and evaluated recommendation set Completed questionnaire Also reviewed and evaluated sets of recommendations from 3 friends each Results Defining Types of Recommendations GOOD: User likes USEFUL Not yet read/viewed Good Recs. (Precision) •% items user felt interested in Useful Recs. •Subset of Good Recs. •User felt interested in and had not read / viewed yet Comparing Human Recommenders to RS: “Good” and “Useful” Recommendations % Good Recommendations % Useful Recommendations 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0 Amazon Sleeper Rating Friends (10) Zone (8) (9) (15) Amazon (15) Reel Movie Friends (5-10) Critic (20) (9) Movies Books Ave. Std. Error (x) No. of Recommendations RS Average However users like online RS Number preferring system Do you prefer recommendations from friends or online systems? System Friends 7.00 6.00 5.00 4.00 3.00 2.00 1.00 0.00 Books Movies This result was supported by post test interviews. Why systems over friends? “Suggested a number of things I hadn’t heard of, interesting matches.” “It was like going to Cody’s—looking at that table up front for new and interesting books.” “Systems can pull from a large database—no one person knows about all the movies I might like.” Items users had “Heard of” before % Heard of 70 60 50 40 30 20 10 0 Amazon S leeper RatingZone Friends Books Amazon MovieCritic Reel Friends Movies Friends recommended mostly “old” previously experienced items What systems did users prefer? Yes Would you use system again? No Amazon Sleeper 0.8 Average Rating 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 Books RatingZone Amazon Reel MovieCritic Movies Sleeper and Amazon books average highest ratings Split opinions on Reel, MovieCritic Why did some systems… Provide useful recommendations but leave users unsatisfied? RatingZone, MovieCritic & Reel Possible Reasons What predicts overall usefulness of a System? 0.6 Correlation 0.5 All correlations are significant at .05 0.4 0.3 0.2 0.1 0 Good Rec. Useful Rec. Trust Adequate Generating Item Rec. Description Ease of Use Previously Enjoyed Items are important: We term these Trust-Generating Items Adequate Item Description & Ease of Use are important Missing from List: Time to Receive Recommendations & No. of Items to Rate not important! A Question of Trust… GOOD: User likes USEFUL Not yet read/viewed TRUST-GENERATING Previously read/viewed Post Test Interviews showed that users “trust” systems if they have already sampled some recommendations •Positive Experiences lead to “trust •Negative Experiences with Recommended Items lead to mistrust of system A Question of Trust … Number of Trust Generating Items Number of Items 6.00 5.00 4.00 3.00 2.00 1.00 .00 Amazon Books Sleeper RatingZone Amazon Reel M ovieCritic Movies Difference between Amazon and Sleeper highlights the fact that there are different kinds of good Recommender Systems Adequate Item Description: The RatingZone Story % Useful Recs. % Useful For Both Versions of RatingZOne 45 40 35 30 25 20 15 0 % of Version 1 and 60% of Version 2 users found item description adequate 10 5 0 Version 1: Without Description Version 2: With Description An adequate item description, and links to other sources about item was a crucial factor in users being convinced by a recommendation. System Transparency Why was this item recommended? Do users understand why an item was recommended % Good Recommendations Effect of System Transparency on Recommendation 60 50 40 30 20 10 0 System Reasoning was Transparent System Reasoning Not Transparent Users mentioned this factor in post test interviews Discussion & Design Recommendations Design Recommendations: Justification Justify your Recommendations Adequate Item Information: Providing enough detail about item for user to make choice System Transparency: Generate (at least some) recommendations which are clearly linked to the rated items Explanation: Provide an Explanation, why the item was recommended. Community Ratings: Provide link to ratings / reviews by other users. If possible, present numerical summary of ratings. Design Recommendations:Accuracy vs. Less Input Don’t sacrifice accuracy for the sake of generating quick recommendations. Users don’t mind rating more items to receive quality recommendations. A possible way to achieve this: have multilevel recommendations. Users can initially use the system by providing one rating, and are offered subsequent opportunities to refine recommendation One needs a happy medium between too little input (leading to low accuracy) and too much input (leading to user impatience) Design Recommendations: New Unexpected Items Users like Rec. Systems as they provide information about new, unexpected items. List of recommended items should include new items which the user might not find out in any other way. List could also include some unexpected items (e.g., from other topics / genres) which the user might not have thought of themselves. Design Recommendations: Trust Generating Items Users (especially first time users) need to develop trust in the system. Trust in system is enhanced by the presence of items that the user has already enjoyed. Generating some very popular (which have probably been experienced previously) in the initial recommendation set might be one way to achieve this. Design Recommendations: Mix of Items Systems need to provide a mix of different kinds of items to cater to different users: Trust Generating Items: A few very popular ones, which the system has high confidence in Unexpected Items: Some unexpected items, whose purpose is to allow users to broaden horizons. Transparent Items: At least some items for which the user can see the clear link between the items he /she rated and the recommendation. New Items: Some items which are new. Question: Should these be presented as a sorted list / unsorted list/ different categories of recommendations? Design Recommendations: Continuous Scales for Input Allow users to provide ratings on a continuous scale. One of the reasons users liked Sleeper was because it allowed them to rate on a continuous scale. Users did not like binary scales. Limitations of Study Simulated first-time visit, did not allow system to learn user preferences over time Source of recommendations known to subjects— might have biased towards friends Fairly homogenous group of subjects, no novice users Future Plans: Second Generation Music Recommender Systems •Have evolved beyond previous systems •Use a variety of sophisticated algorithms to map users preferences over music domain •Require a lot more input from the user •Users can sample recommendations during the study! MusicBudha (Mubu.com): Exploring Genres Mubu.com: Exploring Jazz Styles Mubu.com: Rating Samples Mubu.com: Recommendations as Audio Samples The Turing Test for Music Recommender Systems Compare systems, friends and experts Anonymize the source of recommendation Study Design Friends Music Experts Online RS Goal: Compare recommendations by online RS, Experts (who have same information as RS) & Friends In conclusion… So far we have heard what this study tells us about Recommender Systems But what (if anything) does it have to say about human nature? Recommender Systems tantalize us with the idea that we are not as unique and unpredictable as we think we are. Study results show that Recommender Systems do not know us better than our friends! But … 11 out of 19 users preferred Recommender Systems over Friends Recommendations! Ultimately, we all want to be tables in a database! Email: [email protected] Web address: http://sims.berkeley.edu/~sinha