PAIC 2026 - PTNK AI Challenge 2026

1. Problem Overview The objective of this project is to develop an Automated Essay Scoring (AES) system for IELTS writing tasks using advanced Natural Language Processing (NLP) and Deep Learning techniques. The system aims to help learners quickly and objectively assess their writing proficiency. 2. Training Data The dataset consists of real IELTS writing test essays with the following main components: - Prompt: The topic or question that candidates are required to address. - Image Description: A detailed description of charts or images in Writing Task 1 (currently not used in the model but available in the dataset). - Essay: The written response produced by the candidate. - Overall Score (Label): The final band score assigned by human examiners, ranging from 1.0 to 9.0. - Component scores for Task Response, Lexical Resource, Coherence and Cohesion, and Grammar Range and Accuracy. 3. Problem Requirements Input: An essay, its corresponding prompt, and optionally an image description. Output: A real-valued number representing the predicted band score of the essay. 4. Evaluation Metrics The model’s performance is evaluated based on the agreement between machine-generated scores and human scores: MAE (Mean Absolute Error): The average absolute difference between predicted and human-assigned scores. This is the sole evaluation metric, indicating how far the model’s scores deviate from those of human examiners on average.