Connecting Users to Items Through Tags Shilad Sen Macalester College Jesse Vig John Riedl GroupLens Research Tagommenders Analyze user interactions to infer liking preferences for tag concepts ID: 551266
Download Presentation The PPT/PDF document "Tagommenders" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Tagommenders:Connecting Users to Items Through Tags
Shilad SenMacalester CollegeJesse Vig, John RiedlGroupLens ResearchSlide2Slide3Slide4Slide5Slide6Slide7
Tagommenders
Analyze user interactions to infer liking (preferences) for tag concepts.Recommend items related to tag concepts liked by users.Slide8Slide9
Tagommender GoalsRecommend items using just tags. (Delicious)Improve item recommendations with ratings by by using tags. (LibraryThing
/ Amazon)accuracyflexibilityexplainability (Vig, IUI 2009).Slide10
Tagommender
Flow Chart
WALL-E
animation
robots
pixar
tag preference inference
tag-based recommendationSlide11
MovieLens Tagging
Tagging
introduced in 2006
15,000
distinct tags
127,000
tag applications:
<user, tag, movie>
4000
users applied >= 1 tag
7700 movies
with >= 1 tag appSlide12
OutlineTag preference inferenceItem recommendationAuto-tagging and wrap-upSlide13
OutlineTag preference inferenceItem
recommendationAuto-tagging and wrap-upSlide14
Step 1: Tag Preference Inferenceanimation
robotspixar
?
Infer a user’s interest in tags from:
tags
user
applied
tags
user
searched
for
u
ser’s
clicks
on movie
hyperlinks
user’s
movie
ratingsSlide15
118,017 ratingsby 995 usersSlide16
Preferences for Tags Searched / AppliedSlide17
Movie-rating algorithm
carsSlide18
Movie-Rating Algorithm
cars
4 of 12
1 of 36
9 of 38
0.8
0.1
0.9Slide19
Bayes-Rating AlgorithmGenerative Model:Expressive probabilistic processes.Model movie ratings.Separate model for every user,
tag.Slide20
Jill’s Ratings for animated Movies
N(μ=3.8,σ=0.7)
Bayes
-Rating AlgorithmSlide21
all possible normal dists for ratings for animated movies
WALL-E
not
t
t
= animation
p(t
| WALL-E)
1.0 -
p(t
| WALL-E)
N(μ
u,t
,σ
u,t
)
N(μ
u
,σ
u
)
N(μ=2.0,σ=1.0)
N(μ=4.0,σ=0.5)
Bayes
-Rating AlgorithmSlide22
All movies
m rated by Jill tagged with animation
not
t
t
= animation
Toy Story
WALL-E
Shrek
all possible normal
dists
for ratings for animated movies
Bayes
-Rating AlgorithmSlide23Slide24
OutlineTag preference inferenceItem recommendation
Auto-tagging and wrap-upSlide25
Tagommender
Flow Chart
WALL-E
animation
robots
pixar
tag preference inference
tag-based recommendationSlide26
Step #2: Tag-Based RecommendationStandard machine learning problemWith / without ratingsSix standard recommender baselines
Evaluate predictive performanceSlide27Slide28
OutlineTag preference inference
Item recommendationAuto-tagging and wrap-upSlide29
Inferred pref for girlie movie:
Rating for “Runaway Bride”AliceBob
Mike
(other users)
….
…
cosine similarity
= 0.45
Using Tag Preferences for Tag InferenceSlide30
Top 10 Inferred Tags Not Already Appliedmovietag
cosine simPearl Harbor (2001)disaster0.47Runaway Bride (1999)
girlie movie
0.45
Beauty and the
Beast (1991)
talking animals
0.42
Armageddon (1998)
will smith
0.41
Cinderella (1950)
cartoon
0.40
Inconvenient Truth (2006)
documentary
0.40
The Little Mermaid (1989)
musical
0.40Gone in 60 Seconds (2000)
exciting0.39My Best Friend’s Wedding (1997)chick flick0.39Billy Madison (1995)very funny
0.39Slide31
Summary of TagommendersTag preference inference:Systems can infer user preferences for tags.Item ratings help tag pref
inference.Tag prefs can be used for auto-tagging.Tagommenders outperform traditional recommenders:Without ratings: moderate edge (10%).With ratings: slight edge (2%).Slide32
Future Work
Alternative modalities for tags.Quality vs. preference.Thank You!GroupLens.
MovieLens users.
NSF grants IS 03-24851 and IIS 05-34420.
Macalester College.
Slide33
Shilad Senssen@macalester.edu
(photo by flickr user SantiMB)