Alert button

Meme Sentiment Analysis Enhanced with Multimodal Spatial Encoding and Facial Embedding

Mar 03, 2023

Share this with someone who'll enjoy it:

Internet memes are characterised by the interspersing of text amongst visual elements. State-of-the-art multimodal meme classifiers do not account for the relative positions of these elements across the two modalities, despite the latent meaning associated with where text and visual elements are placed. Against two meme sentiment classification datasets, we systematically show performance gains from incorporating the spatial position of visual objects, faces, and text clusters extracted from memes. In addition, we also present facial embedding as an impactful enhancement to image representation in a multimodal meme classifier. Finally, we show that incorporating this spatial information allows our fully automated approaches to outperform their corresponding baselines that rely on additional human validation of OCR-extracted text.

* In: Longo, L., OReilly, R. (eds) Artificial Intelligence and Cognitive Science. AICS 2022. Communications in Computer and Information Science, vol 1662. Springer, Cham  * Published as chapter in ISBN:978-3-031-26438-2  

Share this with someone who'll enjoy it: