Tag recommendation in social bookmarking sites like deli
1. Tag Recommendation in Social
Bookmarking sites like Deli.cio.us
Varun Ahuja (201206628)
Vinay Singri (201305592)
Tanuj Sharma ( 201101138 )
2. Introduction
ď‚—Automated process of suggesting
relevant keywords given a
dataset
ď‚—Given link L, description D, and
user U, a set of personalized tags
CT(L) are suggested with help
from given dataset.
3. First Approach – STaR ( Social Tag
Recommender System )
Divided in 3 major steps – Pre-processing,
Indexing and Recommendation
Pre-processing – Remove useless tags, Case
Folding, Spam Removal
Indexing – Index existing tags against users.
Recommendation – Combine outputs of Title to
Tag, Resource Profile, User Profile
Recommender.
4. Problems in First Approach
ď‚—Not all tags from the dataset
appeared.
ď‚—Low Precision and Low Recall
ď‚—Without crawling the given link,
this approach gives low accuracy
5. Final Approach – Supervised Learning Model
ď‚—Modelled as a ranking problem of
candidate tags of a given URL
Consists of 3 stages –
â—¦Candidates Tag Extraction
â—¦SVM Features Construction
â—¦Ranking Process
ď‚—Ranking SVM is used for ranking candidate
tags.
6. Candidates Tag Extraction
Extracted from –
â—¦Description field of link L
â—¦Tags assigned by the same user U
previously
â—¦Tags to assigned to the same link L by other
users
ď‚—Given link L, user U, candidate tags
CT{L} = { description(L) union Tags(U) union
Tags(L) }
7. SVM Features Construction
5 features used for each Candidate Tag ( CT ) –
ď‚—Candidate Tag's Term Frequency (TF) in link's description
terms
ď‚—Candidate Tag's Term Frequency (TF) in link's URL terms
Candidate Tag’s Term Frequency (TF) in T{Rj} (tags
assigned to the same URL in the training data).
Candidate Tag’s Term Frequency (TF) in T{Ui} (tags
assigned previously by user in the training data.)
ď‚—Times of candidate tag being assigned as a tag in the
training data.
8. Ranking
ď‚—For any link in test dataset, Candidate
Tags are extracted
ď‚—Features stored for each candidate tag.
ď‚—SVM ranking model ranks the candidate
tags from top to bottom
ď‚—Top K tags selected
10. Future Work
ď‚—Extension to various datasets
ď‚—Giving more enriched
recommendation for the seed URL
ď‚—Candidate Tags can be expanded
using content similarity based KNN
model.
11. References
ď‚— STaR: a Social Tag Recommender System Cataldo
Musto, Fedelucio Narducci, Marco de Gemmis,
Pasquale Lops, and Giovanni Semeraro
Department of Computer Science, University of
Bari, Italy
• Social Tag Prediction Base on
Supervised Ranking Model
Hao Cao, Maoqiang Xie, Lian Xue, Chunhua Liu, Fei
Teng and Yalou Huang
College of Software, Nankai University, Tianjin,
P.R.China