wangkongqiang 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, Data Science & Analytics, Natural Language Processing · Depth: Expert, medium

Summary

The wangkongqiang system, developed for SemEval-2026 Task 9, addresses the detection of multilingual, multicultural, and multievent online polarization across three subtasks: Polarization Detection, Polarization Type Classification, and Manifestation Identification. Focusing on English and Spanish, the system employs google-bert-bertbase-uncased and microsoft-debertav3-base pre-trained language models. Its methodology includes visual analysis of training data, data augmentation via the gemma-3-27b-it generative model using prompts, and training multiple single models with hyperparameter optimization. The system achieved Macro F1 scores of 0.7805 (English) and 0.7155 (Spanish) for Subtask 1, 0.2603 (English) and 0.4647 (Spanish) for Subtask 2, and 0.2766 (English) and 0.3322 (Spanish) for Subtask 3, securing a good ranking in the competition.

Key takeaway

For NLP Engineers developing multilingual text classification systems, particularly for online polarization detection, consider integrating generative models like gemma-3-27b-it for data augmentation. Your approach should include visual data analysis and rigorous hyperparameter tuning of pre-trained models such as BERT and DeBERTa. This strategy can significantly improve Macro F1 scores across diverse languages and cultural contexts, enhancing your system's robustness and competitive performance in complex semantic evaluation tasks.

Key insights

Pre-trained language models combined with generative data augmentation enhance multilingual polarization detection.

Principles

• Visual data analysis guides model development.
• Generative models augment training datasets effectively.
• Hyperparameter tuning optimizes single model performance.

Method

The system visually analyzes training data, augments it using gemma-3-27b-it with prompts, trains multiple single models, and optimizes hyperparameters for test set prediction.

In practice

• Augment text data using gemma-3-27b-it prompts.
• Evaluate bertbase-uncased and debertav3-base models.
• Systematically tune model hyperparameters.

Topics

• SemEval-2026
• Online Polarization Detection
• Multilingual Text Classification
• Data Augmentation
• Generative AI
• Pre-trained Language Models

Best for: AI Engineer, Research Scientist, AI Scientist, Machine Learning Engineer, NLP 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.
• Enhancing SDG-Text Classification with Combinatorial Fusion Analysis and Generative AI — Combinatorial Fusion Analysis (CFA) significantly enhances SDG text classification by combining diverse models and synthetic data.
• Negation-Aware Data Augmentation for Portuguese Natural Language Inference — Targeted negation-aware data augmentation improves BERT-based NLI model performance on Portuguese datasets by reducing negation bias.
• Low-resource Language Discrimination Towards Chinese Dialects with Transfer learning and Data Augmentation — Transfer learning and data augmentation effectively overcome low-resource challenges in Chinese dialect discrimination.
• From Comments to Trends: Studi NLP dan Computational Linguistics dalam Mendeteksi Topik Viral dari… — NLP and computational linguistics are crucial for detecting social media trends from vast, informal, and diverse comment data.
• 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.

Open in AIssential →

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