Skip to content

Latest commit

 

History

History
39 lines (31 loc) · 987 Bytes

README.md

File metadata and controls

39 lines (31 loc) · 987 Bytes

Adapting NER taggers to emerging entities with gazetteers

Thesis by Robin Löbbert, 31.07.2023

Outline

  1. Configuration
    • configs
  2. Search
    • search
  3. Models
    • models
  4. Data Preprocessing
    • data
    • data_preparation
    • data_metrics
    • data_preprocessing
  5. Hyperparameter tuning
    • hyperparameter_tuning
  6. Experiments
    1. experiments > 01_performance
    2. experiments > 02_content
      • data_augmentation
    3. experiments > 03_adaptation_emerging_entities
  7. Evaluation
    • evaluation

frameworks is just a storage directory for code of T5-ASP.

Reproduce results

  • Reproducing the results for T5-ASP, just run run.py in the respective experiment folder.
  • To reproduce results for FLAIR, run models > flair_roberta.py.
  • To reproduce results for DictMatch, run models > dict_match.py.
  • To reproduce results for SearchMatch, run models > search_match.py.

Evaluate results

Run the respective Juypter notebooks after running the experiments.