Named Entity Recognition for FinTech
Financial named entity recognition with LLM-assisted annotation, HMM, and CRF.
Named Entity Recognition for FinTech
This course project studied named entity recognition in financial news. The project combined data collection, LLM-assisted preliminary annotation, human verification, HMM implementation, CRF comparison, and model evaluation.
What It Does
- Builds a labeled financial-news NER dataset.
- Implements an HMM-based sequence labeling model.
- Compares HMM results against a CRF baseline.
- Visualizes scores, transition patterns, and confusion matrices.
My Work
- Served as group leader.
- Led topic selection, prompt screening, financial news collection, data cleaning, and visualization.
- Completed the code implementation, paper writing, presentation slides, diagrams, README, and final defense preparation.
Tech Stack
Python, Jupyter Notebook, PyTorch, NumPy, sklearn-crfsuite, Matplotlib, Seaborn.