Presentation is loading. Please wait.

Presentation is loading. Please wait.

INHA UNIVERSITY INCHEON, KOREA Collaborative Tagging in Recommender Systems AE-TTIE JI 1, CHEOL YEON 1, HEUNG-NAM KIM 1, AND GEUN-SIK.

Similar presentations


Presentation on theme: "INHA UNIVERSITY INCHEON, KOREA Collaborative Tagging in Recommender Systems AE-TTIE JI 1, CHEOL YEON 1, HEUNG-NAM KIM 1, AND GEUN-SIK."— Presentation transcript:

1 INHA UNIVERSITY INCHEON, KOREA Collaborative Tagging in Recommender Systems AE-TTIE JI 1, CHEOL YEON 1, HEUNG-NAM KIM 1, AND GEUN-SIK JO 2 1 Intelligent E-Commerce Systems Laboratory, Department of Computer Science & Information Engineering, Inha University {aerry13, entireboy, aerry13entireboynamiaerry13entireboynami 2 School of Computer Science & Engineering, Inha University, 253 Yonghyun-dong, Incheon, Korea

2 - 2 - INHA UNIVERSITY INCHEON, KOREA 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 - 3 - INHA UNIVERSITY INCHEON, KOREA Introduction

4 - 4 - INHA UNIVERSITY INCHEON, KOREA Collaborative Filtering (CF) Introduction Sparsity Problem Cold-start User Problem

5 - 5 - INHA UNIVERSITY INCHEON, KOREA Collaborative Tagging (CT) Introduction

6 - 6 - INHA UNIVERSITY INCHEON, KOREA Motivation Introduction

7 - 7 - INHA UNIVERSITY INCHEON, KOREA System Architecture Recommender System with CT Part 1: Catching an users latent preference! Part 1: Catching an users latent preference! Candidate Tag Set Generation via CF Part 2: Probabilistic Recommendation! Part 2: Probabilistic Recommendation! Naïve Bayesian Approach

8 - 8 - INHA UNIVERSITY INCHEON, KOREA Matrices Representing Preferences User-item binary matrix, R (r × n) User-item binary matrix, R (r × n) R u,i : whether a user u r prefers an item i n or not. User-tag matrix, A (r × m) User-tag matrix, A (r × m) A u,t : frequency of a tag t m tagged by a user u r. Tag-item matrix, Q (m × n) Tag-item matrix, Q (m × n) Q t,i : frequency of a tag t m for an item i n. Recommender System with CT

9 - 9 - INHA UNIVERSITY INCHEON, KOREA Recommendation Process Recommender System with CT Candidate Tag Set (CTS) Generation via CF Candidate Tag Set (CTS) Generation via CF User-User Similarity Tag Preference

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

11 INHA UNIVERSITY INCHEON, KOREA Case for Data Sparsity Improving limitations of CF via CT

12 INHA UNIVERSITY INCHEON, KOREA Case for Data Sparsity Improving limitations of CF via CT

13 INHA UNIVERSITY INCHEON, KOREA Case for Data Sparsity Improving limitations of CF via CT

14 INHA UNIVERSITY INCHEON, KOREA Case for Cold-start User Improving limitations of CF via CT

15 INHA UNIVERSITY INCHEON, KOREA Recommendation Process Recommender System with CT Item Recommendation Item Recommendation Naïve Bayes Classifier Top-N Items Recommendation

16 INHA UNIVERSITY INCHEON, KOREA Item Recommendation Naïve Bayes Classifier Naïve Bayes Classifier Posterior probability : a preference probability of user u for an item i y with CTS w (u) Prior probability Item-conditional Tag Distribution Top-N Recommendation Top-N Recommendation TopN u items with the highest P u,y, |TopN u | N and TopN u I u = Ø Recommender System with CT

17 INHA UNIVERSITY INCHEON, KOREA Dataset & Evaluation Metric Dataset Dataset (a social bookmarking service) Training data : 21,653 / Testing data : 5,413 Sparsity level of user-item matrix : Evaluation metric Evaluation metric usersItemstags book markings taggings 1,54417,39010,07727,06644,681 Experimental Results

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

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

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

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

22 INHA UNIVERSITY INCHEON, KOREA Conclusion We analyzed the potential of collaborative tagging system for applying to recommendation. 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. 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 INHA UNIVERSITY INCHEON, KOREA Future Work Noise tags can be included in CTS. 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. There are common issues of keyword-based analysis. Polysemy, synonymy and basic level variation. Semantic tagging is an interesting approach to address these issues.


Download ppt "INHA UNIVERSITY INCHEON, KOREA Collaborative Tagging in Recommender Systems AE-TTIE JI 1, CHEOL YEON 1, HEUNG-NAM KIM 1, AND GEUN-SIK."

Similar presentations


Ads by Google