基于二分图的个性化图像标签推荐算法
|
摘要
传统的图像标签推荐方法通过对图像视觉内容的分析计算标签与图像的相关度,完成标签推荐任务.而社会网络图像具有丰富的元数据,例如图像所属群组、地理位置等,充分利用这些元数据对于提高标签推荐的准确性具有积极意义.提出一种基于二分图的个性化图像标签推荐算法,通过充分挖掘图像、群组、地理位置与标签的关系,针对用户提供的少量标签进行个性化图像标签推荐.该算法建立了图像-标签、群组-标签、地理位置-标签等三个二分图模型,考虑到每个标签的重要性不同,引入TF-IDF(Term Frequency-Inverse Document Frenquency)技术对标签进行加权处理.利用二分图将初始标签分值进行信息扩散,计算出最终标签分值向量,并将该向量中分值较高的标签作为推荐结果.实验结果表明,融合了图像与群组、地理位置等元数据的个性化图像标签推荐结果的NDCG(Normalized Discounted Cumulative Gain)值优于仅单方面考虑图像、群组以及地理位置的标签推荐结果.
Traditional image tag recommendation methods mainly concentrated on the analysis of the visual content of the image and the calculation of the correlation between the tag and the image to achieve the tag recommendation task.However,social network images have rich metadata,such as the images' group and site,etc.It is of great importance to make full use of these metadata to improve the accuracy of tag recommendation.In this paper,wepropose a personalized image tag recommendation algorithm based on the bipartite graph model.Furthermore,the proposed algorithm recommends personalized image tags for users with only a small number of user-supplied tags by fully exploiting the relationship between images,groups,sites and tags.In addition,the proposed algorithm constructs three bipartite graph models,such as image-tag model,group-tag model and site-tag model.As the importance of each tag is different,we introduce the TF-IDF(Term Frequency-Inverse Document Frenquency)technology to weight for each tag.The bipartite graph is used to spread the initial tag value,and the final tag value vector is constructed.Then,the tags with higher values in this vector are used as the recommendation results.Exploiting Normalized Discounted Cumulative Gain(NDCG)as performance evaluation criteria,experimental results demonstrate that the NDCG value of the proposed personalized image tag recommendation algorithm which combines three types of metadata information(such as image,group and site)is significantly higher than the NDCG value of the image tag recommendation results which only using image,group and site information individually.
Traditional image tag recommendation methods mainly concentrated on the analysis of the visual content of the image and the calculation of the correlation between the tag and the image to achieve the tag recommendation task.However,social network images have rich metadata,such as the images' group and site,etc.It is of great importance to make full use of these metadata to improve the accuracy of tag recommendation.In this paper,wepropose a personalized image tag recommendation algorithm based on the bipartite graph model.Furthermore,the proposed algorithm recommends personalized image tags for users with only a small number of user-supplied tags by fully exploiting the relationship between images,groups,sites and tags.In addition,the proposed algorithm constructs three bipartite graph models,such as image-tag model,group-tag model and site-tag model.As the importance of each tag is different,we introduce the TF-IDF(Term Frequency-Inverse Document Frenquency)technology to weight for each tag.The bipartite graph is used to spread the initial tag value,and the final tag value vector is constructed.Then,the tags with higher values in this vector are used as the recommendation results.Exploiting Normalized Discounted Cumulative Gain(NDCG)as performance evaluation criteria,experimental results demonstrate that the NDCG value of the proposed personalized image tag recommendation algorithm which combines three types of metadata information(such as image,group and site)is significantly higher than the NDCG value of the image tag recommendation results which only using image,group and site information individually.
引文
[1]Sigurbj9rnsson B,van Zwol R.Flickr tag recommendation based on collective knowledge∥Proceedings of the 17th International Conference on World Wide Web.Beijing,China:ACM,2008:327-336.
[2]Zhang J,Yang Y,Tian Q,et al.Personalized social image recommendation method based on user-image-tag model.IEEE Transactions on Multimedia,2017,19(11):2439-2449.
[3]Schenkel R,Crecelius T,Kacimi M,et al.Efficient top-k querying over social-tagging networks∥Proceedings of the 31st Annual International ACM SIGIR Conference on Research and Development in Information Retrieval.Singapore,Singapore:ACM,2008:523-530.
[4]Zhang Z K,Zhou T,Zhang Y C.Personalized recommendation via integrated diffusion on user-item-tag tripartite graphs.Physica A:Statistical Mechanics and Its Applications,2010,389(1):179-186.
[5]Qian X M,Feng H,Zhao G S,et al.Personalized recommendation combining user interest and social circle.IEEE Transactions on Knowledge&Data Engineering,2014,26(7):1763-1777.
[6]Yang X W,Steck H,Liu Y.Circle-based recommendation in online social networks∥Proceedings of the18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.Beijing,China:ACM,2012:1267-1275.
[7]Phelan O,McCarthy K,Smyth B.Using twitter to recommend real-time topical news∥Proceedings of the 3rd ACM Conference on Recommender Systems.New York,NY,USA:ACM,2009:385-388.
[8]Java A,Song X D,Finin T,et al.Why we twitter:understanding microblogging usage and communities∥Proceedings of the 9th WebKDDand 1st SNA-KDD 2007 Workshop on Web Mining and Social Network Analysis.San Jose,CA,USA:ACM,2007:56-65.
[9]Rafailidis D,Axenopoulos A,Etzold J,et al.Content-based tag propagation and tensor factorization for personalized item recommendation based on social tagging.ACM Transactions on Interactive Intelligent Systems,2014,3(4):Article No.26.
[10]Liu J,Li Z C,Tang J H,et al.Personalized geo-specific tag recommendation for photos on social websites.IEEE Transactions on Multimedia,2014,16(3):588-600.
[11]Li X R,Snoek C G M,Worring M,et al.Fusing concept detection and geo context for visual search∥Proceedings of the 2nd ACMInternational Conference on Multimedia Retrieval.Hong Kong,China:ACM,2012:Article No.4.
[12]Silva A,Martins B.Tag recommendation for georeferenced photos∥Proceedings of the 3rd ACM SIGSPATIAL International Workshop on Location-Based Social Networks.Chicago,IL,USA:ACM,2011:57-64.
[13]Moxley E,Kleban J,Xu J J,et al.Not all tags are created equal:Learning flickr tag semantics for global annotation∥IEEE International Conference on Multimedia and Expo.New York,NY,USA:IEEE,2009:1452-1455.
[14]Kleban J,Moxley E,Xu J J,et al.Global annotation on georeferenced photographs∥Proceedings of the ACM International Conference on Image and Video Retrieval.Santorini,Greece:ACM,2009:Article No.12.
[15]Moxley E,Kleban J,Manjunath B S.Spirittagger:A geo-aware tag suggestion tool mined from flickr∥Proceedings of the 1st ACMInternational Conference on Multimedia Information Retrieval.Vancouver,Canada:ACM,2008:24-30.
[16]Serdyukov P,Murdock V,van Zwol R.Placing flickr photos on a map∥Proceedings of the 32nd International ACM SIGIR Conference on Research and Development in Information Retrieval.Boston,MA,USA:ACM,2009:484-491.
[17]Geng X,Zhang H W,Bian J W,et al.Learning image and user features for recommendation in social networks∥IEEE International Conference on Computer Vision.Santiago,Chile:IEEE,2015:4274-4282.
[18]Zhou T,Ren J,Medo M,et al.Bipartite network projection and personal recommendation.Physical Review E,2007,76(2):046115.
[19]Shang M S,Zhang Z K,Zhou T,et al.Collaborative filtering with diffusion-based similarity on tripartite graphs.Physica A:Statistical Mechanics and Its Applications,2010,389(6):1259-1264.
[20]Liu D,Hua X S,Yang L J,et al.Tag ranking∥Proceedings of the 18th International Conference on World Wide Web.Madrid,Spain:ACM,2009:351-360.
[21]Li X R,Snoek C G M,Worring M.Learning tag relevance by neighbor voting for social image retrieval∥Proceedings of the 1st ACM SIGMMInternational Conference on Multimedia Information Retrieval.Vancouver,Canada:ACM,2008:180-187.
[2]Zhang J,Yang Y,Tian Q,et al.Personalized social image recommendation method based on user-image-tag model.IEEE Transactions on Multimedia,2017,19(11):2439-2449.
[3]Schenkel R,Crecelius T,Kacimi M,et al.Efficient top-k querying over social-tagging networks∥Proceedings of the 31st Annual International ACM SIGIR Conference on Research and Development in Information Retrieval.Singapore,Singapore:ACM,2008:523-530.
[4]Zhang Z K,Zhou T,Zhang Y C.Personalized recommendation via integrated diffusion on user-item-tag tripartite graphs.Physica A:Statistical Mechanics and Its Applications,2010,389(1):179-186.
[5]Qian X M,Feng H,Zhao G S,et al.Personalized recommendation combining user interest and social circle.IEEE Transactions on Knowledge&Data Engineering,2014,26(7):1763-1777.
[6]Yang X W,Steck H,Liu Y.Circle-based recommendation in online social networks∥Proceedings of the18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.Beijing,China:ACM,2012:1267-1275.
[7]Phelan O,McCarthy K,Smyth B.Using twitter to recommend real-time topical news∥Proceedings of the 3rd ACM Conference on Recommender Systems.New York,NY,USA:ACM,2009:385-388.
[8]Java A,Song X D,Finin T,et al.Why we twitter:understanding microblogging usage and communities∥Proceedings of the 9th WebKDDand 1st SNA-KDD 2007 Workshop on Web Mining and Social Network Analysis.San Jose,CA,USA:ACM,2007:56-65.
[9]Rafailidis D,Axenopoulos A,Etzold J,et al.Content-based tag propagation and tensor factorization for personalized item recommendation based on social tagging.ACM Transactions on Interactive Intelligent Systems,2014,3(4):Article No.26.
[10]Liu J,Li Z C,Tang J H,et al.Personalized geo-specific tag recommendation for photos on social websites.IEEE Transactions on Multimedia,2014,16(3):588-600.
[11]Li X R,Snoek C G M,Worring M,et al.Fusing concept detection and geo context for visual search∥Proceedings of the 2nd ACMInternational Conference on Multimedia Retrieval.Hong Kong,China:ACM,2012:Article No.4.
[12]Silva A,Martins B.Tag recommendation for georeferenced photos∥Proceedings of the 3rd ACM SIGSPATIAL International Workshop on Location-Based Social Networks.Chicago,IL,USA:ACM,2011:57-64.
[13]Moxley E,Kleban J,Xu J J,et al.Not all tags are created equal:Learning flickr tag semantics for global annotation∥IEEE International Conference on Multimedia and Expo.New York,NY,USA:IEEE,2009:1452-1455.
[14]Kleban J,Moxley E,Xu J J,et al.Global annotation on georeferenced photographs∥Proceedings of the ACM International Conference on Image and Video Retrieval.Santorini,Greece:ACM,2009:Article No.12.
[15]Moxley E,Kleban J,Manjunath B S.Spirittagger:A geo-aware tag suggestion tool mined from flickr∥Proceedings of the 1st ACMInternational Conference on Multimedia Information Retrieval.Vancouver,Canada:ACM,2008:24-30.
[16]Serdyukov P,Murdock V,van Zwol R.Placing flickr photos on a map∥Proceedings of the 32nd International ACM SIGIR Conference on Research and Development in Information Retrieval.Boston,MA,USA:ACM,2009:484-491.
[17]Geng X,Zhang H W,Bian J W,et al.Learning image and user features for recommendation in social networks∥IEEE International Conference on Computer Vision.Santiago,Chile:IEEE,2015:4274-4282.
[18]Zhou T,Ren J,Medo M,et al.Bipartite network projection and personal recommendation.Physical Review E,2007,76(2):046115.
[19]Shang M S,Zhang Z K,Zhou T,et al.Collaborative filtering with diffusion-based similarity on tripartite graphs.Physica A:Statistical Mechanics and Its Applications,2010,389(6):1259-1264.
[20]Liu D,Hua X S,Yang L J,et al.Tag ranking∥Proceedings of the 18th International Conference on World Wide Web.Madrid,Spain:ACM,2009:351-360.
[21]Li X R,Snoek C G M,Worring M.Learning tag relevance by neighbor voting for social image retrieval∥Proceedings of the 1st ACM SIGMMInternational Conference on Multimedia Information Retrieval.Vancouver,Canada:ACM,2008:180-187.