A Cross-Modal Approach for Extracting Semantic Relationships Between Concepts Using Tagged Images

Comparative analysis on cross-modal information retrieval: A review

“…Comparison based on two new evaluation metrics and recent image and text features is also incorporated in the new work. [59] has proposed a crossmodal technique for extracting semantic relationship between classes using annotated images. Firstly, both visual features and text are projected onto a latent space using CCA, and then the probabilistic interpretation of CCA is utilized for calculating the representative distribution of the latent variable for each class.…”

Survey on visual sentiment analysis

“…post message, tags and comments) by exploiting common Sentiment Analysis systems that works on textual contents [29, 127], and try to learn ML systems able to infer that polarity from the associated visual content. These techniques have achieved interesting improvements in the tasks of image content recognition, automatic annotation and image retrieval [87, 128–133]. However, it is impossible to know if such user provided text is related to the image content or to the sentiment it conveys.…”

[Invited Paper] A Review of Web Image Mining

“…The latent topic based visual language models are learned with the images gathered from Flickr. Katsurai et al 60) proposed a cross-modal approach based on probabilistic CCA 3) to measure distance between concepts using a large number of tagged photos.…”