Researchers in China say they created the I LIKE ANIME VIDEO GAMES MAYBE 3 PEOPLE Funny Gamer Sarcasm T-Shirt ironic detection AI that achieves the most advanced performance on a data set pulled from Twitter. AI uses multimodal learning that combines text and images as both are often needed to understand whether a person is being sarcastic.
I LIKE ANIME VIDEO GAMES MAYBE 3 PEOPLE Funny Gamer Sarcasm T-Shirt, hoodie, tank top, sweater
The discovery of sarcasm can aid in analyzing emotions and understanding from public sources about public attitudes on a particular topic, the researchers argue. In a challenge launched earlier this year, Facebook is using multimodal AI to figure out if memes are violating its terms of service.
The researchers’ AI focuses on the I LIKE ANIME VIDEO GAMES MAYBE 3 PEOPLE Funny Gamer Sarcasm T-Shirt difference between text and images, then combines those results to make predictions. It also compares hashtags with tweet text to help gauge the sentiment a user is trying to convey.
“In particular, the input tokens will give high attention to areas of images that contradict them, since inconsistencies are a key feature of irony,” the article wrote. “Since a mismatch can appear only in the text (eg a satirical text combined with an unrelated image), internal modal mismatches need to be considered. ”
On the data set taken from Twitter, the I LIKE ANIME VIDEO GAMES MAYBE 3 PEOPLE Funny Gamer Sarcasm T-Shirt model achieved a 2.74% improvement in ironic detection F1 scores compared to HFM, a multimodal detection model introduced last year. The new model also achieved 86% accuracy, compared with 83% for HFM.
The paper is published by the Chinese Academy of Sciences and the Institute of Information Engineering, both in Beijing, China. The paper was presented this week at the conference on Virtual Experimental Methods in Natural Language Processing (EMNLP).
AI is the latest example of multimodal ironic detection to emerge since AI researchers began studying satirical mode in multimodal content on Instagram, Tumblr, and Twitter in 2016.
Researchers from the University of Michigan and the University of Singapore used computer vision and language models to detect irony in television shows, a model detailed in the article. titled “Towards Extreme Multimodal Discovery (A Perfect Paper). That work was marked as part of the Computational Linguistics Association (ACL) last year.