🌌 Chimera: A Cognitive & Aesthetic Sentiment Causality Understanding Framework

This repository provides the source code and datasets for our paper:
📄 Exploring Cognitive and Aesthetic Causality for Multimodal Aspect-Based Sentiment Analysis

---

🚀 Overview

Chimera is a unified framework designed to model cognitive reasoning and aesthetic perception in multimodal aspect-based sentiment analysis (MABSA). It integrates text–image interaction, sentiment causality understanding, and fine-grained multimodal reasoning.

---

📦 Environment Requirements

We recommend using conda for environment management:

conda env create -f Chimera.yaml

---

📚 Datasets

1. Constructed datasets: Twitter2015 (twitter2015), Twitter2017 (twitter2017) and Political Twitter (political_twitter).
2. Image features can be downloaded from Google Drive. Place the downloaded files in the directories data/twitter2015 and data/twitter2017, respectively. For the political_twitter dataset, move the contents of data/twitter2015 and data/twitter2017 into data/political_twitter and extract the two .zip files into the same directory.

---

🤖 Pretrained Model

• The Flan-T5 model is utilized as the backbone. Download the pre-trained model google/flan-t5-base and save it in the directory pretrained/flan-t5-base.

---

🏋️ Training & Evaluation

python run_chimera_15.py

python run_chimera_17.py

python run_chimera_political.py

---

📖 Citation

If you find this repository helpful, please consider starring ⭐ the repo and citing our related work:

@article{xiao2025exploring,
  title={Exploring Cognitive and Aesthetic Causality for Multimodal Aspect-Based Sentiment Analysis},
  author={Xiao, Luwei and Mao, Rui and Zhao, Shuai and Lin, Qika and Jia, Yanhao and He, Liang and Cambria, Erik},
  journal={IEEE Transactions on Affective Computing},
  year={2025},
  publisher={IEEE}
}

@article{xiao2024atlantis,
  title={Atlantis: Aesthetic-oriented multiple granularities fusion network for joint multimodal aspect-based sentiment analysis},
  author={Xiao, Luwei and Wu, Xingjiao and Xu, Junjie and Li, Weijie and Jin, Cheng and He, Liang},
  journal={Information Fusion},
  volume={106},
  pages={102304},
  year={2024},
  publisher={Elsevier}
}

@inproceedings{xiao2024vanessa,
  title={Vanessa: Visual connotation and aesthetic attributes understanding network for multimodal aspect-based sentiment analysis},
  author={Xiao, Luwei and Mao, Rui and Zhang, Xulang and He, Liang and Cambria, Erik},
  booktitle={Findings of the Association for Computational Linguistics: EMNLP 2024},
  pages={11486--11500},
  year={2024}
}

🙏 Acknowledgements

This work is primarily built upon the repositories of MDCA and LAPS. We extend our sincere gratitude to all contributors for their valuable insights and support.

---

✨ Thank you for your interest in Chimera! Feel free to open issues or pull requests to help improve this project.