arXiv preprint

LLMPR: A Novel LLM-Driven Transfer Learning based Petition Ranking Model

LLMPR combines language-model embeddings, transfer learning, and ranking models to prioritize petitions from semantically rich text.

Petition textEmbedding ensembleTransfer-learning modelRanked petitions

Plain-language summary

Large collections of petitions can be difficult to review and prioritize manually. LLMPR represents each petition using several language-model embeddings and uses transfer-learning and ranking models to identify semantically important cases.

Research contribution

  • Comparison and combination of multiple transformer embedding families.
  • Transfer-learning features with classical and ensemble prediction models.
  • Evaluation using representation-quality and ranking metrics.

Publication status

Public preprint available as arXiv:2505.21689.

Code and dataset

Code and dataset links will be added when a public release is available.

Recommended citation

Gayen, A., Chakraborty, S., Sen, M., Paul, S., and Jana, A. (2025). LLMPR: A Novel LLM-Driven Transfer Learning based Petition Ranking Model. arXiv:2505.21689.

BibTeX

@article{gayen2025llmpr, title={LLMPR: A Novel LLM-Driven Transfer Learning based Petition Ranking Model}, author={Gayen, Avijit and Chakraborty, Somyajit and Sen, M. and Paul, S. and Jana, Angshuman}, journal={arXiv preprint arXiv:2505.21689}, year={2025} }

Related work

LLMPR project page · CenterDistill · PacePal