Over the last few years, the social media site Flickr has gained massive popularity. Besides traditional operations on photo sharing, Flickr also offers millions of groups for users to join in order to share photos with others, and the number of groups still increases daily. Choosing among so many options is challenging for users. As such, helping users easily find their desirable groups has become increasingly important. In this paper, we provide a systematic experimental evaluation of several collaborative filtering algorithms to recommend groups for Flickr users. In particular, we design and compare seven Flickr group recommendation models: three memory-based models and four model-based models. Our results suggest that model-based approaches are beneficial compared with memory-based approaches in terms of top-
k
recommendation metric. Models with tags perform well for sparse data, whereas models without tags are more suitable for dense data. Furthermore, incorporating tags in the recommendation algorithms leads to an improvement of precision on the top 2\% performance.
%0 Journal Article
%1 nan_zheng_which_2010
%A Zheng, Nan
%A Li, Qiudan
%A Liao, Shengcai
%A Zhang, Leiming
%D 2010
%J Journal of Information Science
%K Folksonomies, flickr, image, social\_bookmarking, web\_2.0
%N 6
%P 733 --750
%R 10.1177/0165551510386164
%T Which photo groups should I choose? A comparative study of recommendation algorithms in Flickr
%U http://jis.sagepub.com/content/36/6/733.abstract
%V 36
%X Over the last few years, the social media site Flickr has gained massive popularity. Besides traditional operations on photo sharing, Flickr also offers millions of groups for users to join in order to share photos with others, and the number of groups still increases daily. Choosing among so many options is challenging for users. As such, helping users easily find their desirable groups has become increasingly important. In this paper, we provide a systematic experimental evaluation of several collaborative filtering algorithms to recommend groups for Flickr users. In particular, we design and compare seven Flickr group recommendation models: three memory-based models and four model-based models. Our results suggest that model-based approaches are beneficial compared with memory-based approaches in terms of top-
k
recommendation metric. Models with tags perform well for sparse data, whereas models without tags are more suitable for dense data. Furthermore, incorporating tags in the recommendation algorithms leads to an improvement of precision on the top 2\% performance.
@article{nan_zheng_which_2010,
abstract = {Over the last few years, the social media site Flickr has gained massive popularity. Besides traditional operations on photo sharing, Flickr also offers millions of groups for users to join in order to share photos with others, and the number of groups still increases daily. Choosing among so many options is challenging for users. As such, helping users easily find their desirable groups has become increasingly important. In this paper, we provide a systematic experimental evaluation of several collaborative filtering algorithms to recommend groups for Flickr users. In particular, we design and compare seven Flickr group recommendation models: three memory-based models and four model-based models. Our results suggest that model-based approaches are beneficial compared with memory-based approaches in terms of top-
k
recommendation metric. Models with tags perform well for sparse data, whereas models without tags are more suitable for dense data. Furthermore, incorporating tags in the recommendation algorithms leads to an improvement of precision on the top 2\% performance.},
added-at = {2011-03-24T16:45:37.000+0100},
author = {Zheng, Nan and Li, Qiudan and Liao, Shengcai and Zhang, Leiming},
biburl = {https://www.bibsonomy.org/bibtex/2534a1ca0d249a0f0f4300c8323e7b600/boudry},
doi = {10.1177/0165551510386164},
interhash = {cba020b935af4c64956bd3d4a655393f},
intrahash = {534a1ca0d249a0f0f4300c8323e7b600},
journal = {Journal of Information Science},
keywords = {Folksonomies, flickr, image, social\_bookmarking, web\_2.0},
month = dec,
number = 6,
pages = {733 --750},
shorttitle = {Which photo groups should I choose?},
timestamp = {2011-03-24T16:45:40.000+0100},
title = {Which photo groups should I choose? A comparative study of recommendation algorithms in Flickr},
url = {http://jis.sagepub.com/content/36/6/733.abstract},
volume = 36,
year = 2010
}