1. Collaborative Tagging in Recommender Systems AE-TTIE JI1, CHEOL YEON1, HEUNG-NAM KIM1, AND GEUN-SIK JO2
1 Intelligent E-Commerce Systems Laboratory,
Department of Computer Science & Information Engineering, Inha University
{aerry13, entireboy,
2 School of Computer Science & Engineering, Inha University,
253 Yonghyun-dong, Incheon, Korea

2. Recommender System with Collaborative Tagging
Introduction
Recommender System with Collaborative Tagging
Experimental Results
Conclusions
Future Works
Introduction
Recommender System with Collaborative Tagging
Experimental Results
Conclusions
Future Works
Introduction
Recommender System with Collaborative Tagging
Experimental Results
Conclusions
Future Works
Introduction
Recommender System with Collaborative Tagging
Experimental Results
Conclusions
Future Works
Introduction
Recommender System with Collaborative Tagging
Experimental Results
Conclusions
Future Works
Introduction
Recommender System with Collaborative Tagging
Experimental Results
Conclusions
Future Works

3. Introduction

4. Collaborative Filtering (CF)
Introduction
Collaborative Filtering (CF)
Cold-start
User
Problem
Sparsity
Problem

5. Collaborative Tagging (CT)
Introduction
Collaborative Tagging (CT)

6. Introduction
Motivation

7. System Architecture Part 1: Catching an user’s latent preference!
Recommender System with CT
System Architecture
Part 1: Catching an user’s latent preference!
Candidate Tag Set Generation via CF
Part 2: Probabilistic Recommendation!
Naïve Bayesian Approach

8. Matrices Representing Preferences
Recommender System with CT
Matrices Representing Preferences
User-item binary matrix, R (r × n)
Ru,i : whether a user ur prefers an item in or not.
User-tag matrix, A (r × m)
Au,t : frequency of a tag tm tagged by a user ur.
Tag-item matrix, Q (m × n)
Qt,i : frequency of a tag tm for an item in.

9. Recommendation Process
Recommender System with CT
Recommendation Process
Candidate Tag Set (CTS) Generation via CF
User-User Similarity
Tag Preference

10. CTS Generation via CF CTS (Candidate Tag Set) User-user Similarity
Recommender System with CT
CTS Generation via CF
CTS (Candidate Tag Set)
The latent preference of a target user
User-user Similarity
To find k nearest neighbors (KNN) of a target user based on user-tag matrix A
Tag Preference

11. Improving limitations of CF via CT
Case for Data Sparsity

12. Improving limitations of CF via CT
Case for Data Sparsity

13. Improving limitations of CF via CT
Case for Data Sparsity

14. Case for Cold-start User
Improving limitations of CF via CT
Case for Cold-start User

15. Recommendation Process
Recommender System with CT
Recommendation Process
Item Recommendation
Naïve Bayes Classifier
Top-N Items Recommendation

16. Item Recommendation Naïve Bayes Classifier Top-N Recommendation
Recommender System with CT
Item Recommendation
Naïve Bayes Classifier
Posterior probability : a preference probability of user u for an item iy with CTSw(u)
Prior probability
Item-conditional Tag Distribution
Top-N Recommendation
TopNu items with the highest Pu,y , |TopNu| ≤ N and TopNu ∩ Iu = Ø

17. Dataset & Evaluation Metric
Experimental Results
Dataset & Evaluation Metric
Dataset
(a social bookmarking service)
Training data : 21,653 / Testing data : 5,413
Sparsity level of user-item matrix :
Evaluation metric
users
Items
tags
book markings
taggings
1,544
17,390
10,077
27,066
44,681

18. Experimental Results
Benchmark Algorithms
User-based Collaborative Filtering (Badrul Sarwar, and et al., 2000)
Item-based Collaborative Filtering (Mukund Deshpande, and et al., 2004)
KNN size was set to 50 where the performance increase rates were diminished for main comparison.
Recommendation size N = 10

19. Experiments with CTS size
Experimental Results
Experiments with CTS size
The size of CTS, w, can be a significant factor affecting the quality of recommendation.
w was set to 70, which obtained the best quality for main comparisons.
Neighbor size k = 50
Recommendation size N = 10

20. Comparisons of Overall Performance
Experimental Results
Comparisons of Overall Performance
Sparsity of the collected dataset affected the performances of all three methods.
Even though the number of recommended items were small, our method outperformed the other two methods.
Neighbor size k = 50
CTS size w = 70

21. Comparisons for Cold-start User
Experimental Results
Comparisons for Cold-start User
For cold-start users who do not have enough preference information, our method outperformed the other two methods.
Neighbor size k = 50
CTS size w = 70
Recommendation size N = 10

22. Conclusion
We analyzed the potential of collaborative tagging system for applying to recommendation.
User-created tags imply users’ preferences about items as well as metadata about them.
Using tags can partially improve data sparsity and cold-start user problem which are serious limitations of CF recommendation.
Also proposed is a novel recommender system based on collaborative tags of users using CF scheme.
Our algorithm obtained better recommendation quality compared to traditional CF schemes.
It provided more suitable items for user preferences even though the number of recommended items were small.

23. Future Work “Noise” tags can be included in CTS.
Some tags are too personalized or content-criticizable (e.g., bad, myWork, to read etc.)
They should be treated for more personalized and valuable analysis.
There are common issues of keyword-based analysis.
Polysemy, synonymy and basic level variation.
Semantic tagging is an interesting approach to address these issues.