wangkongqiang@EEUCA 2026: Multimodal Identification of Vaccine Critical Content on Social Media

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Intermediate, quick

Summary

A team led by Kongqiang Wang participated in the Multimodal Identification of Vaccine Critical Content on Social Media shared task at EEUCA with ACL 2026. Their objective was to classify vaccine-related memes into "Vaccine critical" (0), "Neutral" (1), or "Pro-vaccine" (2) categories. The task utilized the VaxMeme dataset, which comprises over 10,000 manually annotated vaccination-related memes. The team employed a supervised learning approach, fine-tuning pre-trained Large Language Models (LLMs) such as Qwen2 and Llama series LLMs using Llama-Factory. Their best result on the test set was achieved with a fine-tuned qwen2_1.5B LLM, yielding a Macro F1 score of 0.8153, Accuracy of 0.8185, Precision (Macro) of 0.8151, and Recall (Macro) of 0.8159, securing 12th place among all participating teams.

Key takeaway

For machine learning engineers developing content classification systems, particularly for social media, you should consider fine-tuning specific LLMs like Qwen2_1.5B. This approach demonstrated strong performance in identifying vaccine-critical content from multimodal memes. Explore multimodal datasets like VaxMeme and tools such as Llama-Factory to enhance your model's accuracy and contextual understanding for nuanced content moderation.

Key insights

Fine-tuning LLMs on multimodal data effectively classifies vaccine-related social media content.

Method

The team used supervised learning to fine-tune Qwen2 and Llama series LLMs via Llama-Factory on the VaxMeme dataset for multimodal vaccine-critical meme classification.

In practice

• Fine-tuning Qwen2_1.5B LLM achieved a Macro F1 score of 0.8153.
• Llama-Factory can facilitate LLM fine-tuning for specific classification tasks.
• The VaxMeme dataset supports multimodal vaccine-critical meme detection.

Topics

• Multimodal Classification
• Vaccine Misinformation
• Social Media Content
• Large Language Models
• LLM Fine-tuning
• Qwen2 LLM
• VaxMeme Dataset

Best for: NLP Engineer, AI Scientist, Machine Learning Engineer, Research Scientist

Related on AIssential

• I created an LLM trained solely on Jeffrey Epsteins emails to see how messed up it becomes :) — Fine-tuning an LLM on a controversial, unannotated dataset can reveal unique behavioral shifts and spark technical debate.
• A multimodal large language model for materials science — MatterChat unifies material structure and text using a modular LLM, improving property prediction and scientific reasoning in materials science.
• Scikit-LLM vs. Traditional Text Classifiers: When Should You Use an LLM? — LLMs offer superior zero-shot text classification accuracy and competitive latency, especially with limited data.
• MILM: Large Language Models for Multimodal Irregular Time Series with Informative Sampling — LLMs can effectively process multimodal irregular time series by encoding sampling patterns and values in XML.
• Addressing Data Scarcity in Bangla Fake News Detection: An LLM-Based Dataset Augmentation Approach — LLM-based data augmentation significantly improves fake news detection in low-resource languages by generating diverse synthetic data.
• Fine-Tuning Qwen3-VL — Fine-tuning Qwen3-VL for image-to-HTML requires careful dataset preparation and context length management.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.