MMDF-LDA: An improved Multi-Modal Latent Dirichlet Allocation model for social image annotation

作者：

Highlights：

• A multi-modal data fusion model for social images annotation is proposed.

• A probability topic model is learned by fusing multi-modal metadata.

• Geographical topics are generated from geographical region of social images.

• Patches of social images are annotated by the proposed model.

• Experiments demonstrate the effectiveness of the proposed solution.

摘要

•A multi-modal data fusion model for social images annotation is proposed.•A probability topic model is learned by fusing multi-modal metadata.•Geographical topics are generated from geographical region of social images.•Patches of social images are annotated by the proposed model.•Experiments demonstrate the effectiveness of the proposed solution.

论文关键词：Social image,Multi-modal data fusion,LDA model,Semantic annotation,Geographical topic

论文评审过程：Received 19 November 2017, Revised 7 February 2018, Accepted 10 March 2018, Available online 12 March 2018, Version of Record 30 March 2018.

论文官网地址：https://doi.org/10.1016/j.eswa.2018.03.014