zhangpeng at SemEval-2026 Task 9: Detecting Multilingual, Multicultural and Multievent Online Polarization

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Natural Language Processing · Depth: Advanced, medium

Summary

The "zhangpeng" system, developed by Zhang Peng and Lu Gehao for SemEval-2026 Task 9, addresses the challenge of detecting multilingual, multicultural, and multievent online polarization across three subtasks. For Subtask 1, "Polarization Detection," the system explored classical text representation methods, with the Bag-of-Centroids model achieving the best performance. For Subtask 2, "Polarization Type Classification," and Subtask 3, "Manifestation Identification," the team fine-tuned four pre-trained language models: google-bert, FacebookAI-roberta, dccuchile-bert, and distilbert-multi. The methodology included visual analysis of training data, training multiple single models, and employing voting weight ensemble learning with hyperparameter tuning. On the official test set, the system achieved Macro-F1 scores of 0.6882 (EN) and 0.6711 (SP) for Subtask 1, 0.3752 (EN) and 0.6386 (SP) for Subtask 2, and 0.3561 (EN) and 0.4366 (SP) for Subtask 3.

Key takeaway

For NLP Engineers developing multilingual polarization detection systems, consider a hybrid approach combining classical text representation with fine-tuned pre-trained language models. You should prioritize ensemble learning and rigorous hyperparameter tuning. This optimizes Macro-F1 scores across diverse subtasks, including polarization detection, type classification, and manifestation identification. This strategy can enhance your system's robustness and performance in complex, multicultural online environments.

Key insights

Ensemble learning with classical and fine-tuned LLMs effectively detects multilingual online polarization.

Principles

• Combine classical and deep learning models.
• Ensemble learning improves prediction robustness.
• Hyperparameter tuning is crucial for integrated models.

Method

The system visually analyzes training data, trains multiple single models, and applies voting weight ensemble learning, followed by hyperparameter optimization for the integrated model.

In practice

• Use Bag-of-Centroids for initial text classification.
• Fine-tune BERT-based models for complex subtasks.
• Implement voting ensembles for improved F1 scores.

Topics

• Online Polarization Detection
• Multilingual Text Classification
• Ensemble Learning
• Pre-trained Language Models
• SemEval-2026
• Natural Language Processing

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

Related on AIssential

• mdok-style at SemEval-2026 Task 9: Finetuning LLMs for Multilingual Polarization Detection — Finetuning mid-sized LLMs with QLoRA and augmented data improves multilingual online polarization detection.
• TeamHerald@CHIPSAL 2026: Hate Speech Detection and Sentiment Analysis of Nepali Memes using Transformer-based Architectures and Ensemble Learning — Ensemble strategies for meme analysis perform differently for binary vs. multi-class tasks, requiring task-specific aggregation.
• Multi-Stage Training for Abusive Comment Detection in Indic Languages — Multi-stage training with language preprocessing and model ensembles improves abusive comment detection while minimizing false positives.
• Comparison of Modern Multilingual Text Embedding Techniques for Hate Speech Detection Task — Supervised models with multilingual embeddings and gradient boosted trees excel at hate speech detection.
• FLiP: Towards understanding and interpreting multimodal multilingual sentence embeddings — FLiP models effectively interpret sentence embeddings by linearly extracting lexical content, revealing underlying biases.
• From Logistic Regression to GPT-2: Building a Complete Spam Detection & Sentiment Analysis Pipeline — For imbalanced text classification, prioritize F1-score and Precision-Recall AUC over accuracy and ROC-AUC.

Open in AIssential →

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