This project explores how agentic AI workflows can improve automatic text simplification. Traditional approaches typically rely on a single model or prompt to transform complex text into a simpler version. However, such one-step methods often produce inconsistent results, failing to adequately balance readability, fluency, and meaning preservation.

In this project, the student will design and implement a multi-step (“agentic”) pipeline, where different components (agents) perform specialized roles in the simplification process. The key idea is to decompose the task into smaller, interpretable steps—such as analyzing complexity, rewriting the text, and refining the output—so that the system produces more controlled and higher-quality simplifications.

References

1. Färber, M., Aghdam, P., Im, K., Tawfelis, M. and Ghoshal, H., 2025, April. Simplifymytext: An llm-based system for inclusive plain language text simplification. In European Conference on Information Retrieval (pp. 418-424). Cham: Springer Nature Switzerland. https://link.springer.com/chapter/10.1007/978-3-031-88717-8_32
2. Smirnova, A., Chun, K.B., Rothman, W.L. and Sarma, S., 2025, April. Text simplification for children: Evaluating llms vis-à-vis human experts. In Proceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems (pp. 1-10). https://dl.acm.org/doi/full/10.1145/3706599.3719889
3. Ali, S.M., Sajid, H., Aijaz, O., Waheed, O., Alvi, F. and Samad, A., 2024. Team Sharingans at SimpleText: Fine-tuned LLM based approach to Scientific Text Simplification. In CLEF (Working Notes) (pp. 3174-3181). https://ceur-ws.org/Vol-3740/paper-309.pdf
4. Wu, Q., Bansal, G., Zhang, J., Wu, Y., Li, B., Zhu, E., Jiang, L., Zhang, X., Zhang, S., Liu, J. and Awadallah, A.H., 2024, August. Autogen: Enabling next-gen LLM applications via multi-agent conversations. In First conference on language modeling. https://www.microsoft.com/en-us/research/wp-content/uploads/2023/08/LLM_agent.pdf