svercoutere/legal-bert-base-classifier-refinement-abb

Model Card for Model ID

Legal-BERT Base Entity Classifier

Overview

A fine-tuned BERT-based model for classifying legal entities (such as locations and dates) within the context of legal decision texts. The model is based on nlpaueb/legal-bert-base-uncased and is trained to predict the type of a marked entity span, given its context, using special entity markers [E] ... [/E].

Model Details

• Model Name: legal-bert-base-classifier-refinement-abb
• Architecture: BERT (nlpaueb/legal-bert-base-uncased)
• Task: Entity Classification (NER-style, entity-in-context classification)
• Framework: PyTorch, Hugging Face Transformers
• Author: S. Vercoutere

Intended Use

• Purpose: Automatic classification of legal entities (e.g., location, date) in municipal or governmental decision documents.
• Not Intended For: General-purpose NER, non-legal domains, or tasks outside entity classification.

Training Data

• Source: Annotated legal decision texts from Ghent/Freiburg/Bamberg.
• Entity Types:
  • Locations: impact_location, context_location
  • Dates: publication_date, session_date, entry_date, expiry_date, legal_date, context_date, validity_period, context_period
• Preprocessing:
  • XML-like tags in text, with entities wrapped in <entity_type>...</entity_type>.
  • For training, one entity per sample is marked with [E] ... [/E] in context.
  • Dataset balanced to max 5000 samples per label.

Training Procedure

• Model: nlpaueb/legal-bert-base-uncased
• Tokenization: Hugging Face AutoTokenizer, with [E] and [/E] as additional special tokens.
• Max Sequence Length: 512
• Batch Size: 4
• Optimizer: AdamW
• Learning Rate: 2e-5
• Epochs: 10
• Mixed Precision: Yes (AMP)
• Validation Split: 20%
• Evaluation Metrics: Accuracy, F1, confusion matrix

Evaluation

Validation Accuracy: 0.8454 (on held-out validation set)

Detailed Entity-Level Evaluation:

Entity Label	Precision	Recall	F1-score	Support
context_date	0.7298	0.7867	0.7572	975
context_location	0.8623	0.8992	0.8804	843
context_period	0.6443	0.7007	0.6713	137
entry_date	0.8806	0.6860	0.7712	484
expiry_date	0.7826	0.5180	0.6234	139
impact_location	0.7616	0.8205	0.7900	997
legal_date	0.9685	0.9777	0.9731	943
publication_date	0.9626	1.0000	0.9809	386
session_date	0.9680	0.9597	0.9638	347
validity_period	0.8772	0.7495	0.8083	467
accuracy			0.8454	5718
macro avg	0.8438	0.8098	0.8220	5718
weighted avg	0.8485	0.8454	0.8444	5718

Usage Example

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

tokenizer = AutoTokenizer.from_pretrained("svercoutere/legal-bert-base-classifier-refinement-abb")
model = AutoModelForSequenceClassification.from_pretrained("svercoutere/legal-bert-base-classifier-refinement-abb")

def classify_entity(entity_text, context_text):
    marked_text = context_text.replace(entity_text, f"[E] {entity_text} [/E]", 1)
    inputs = tokenizer(marked_text, return_tensors="pt", truncation=True, max_length=512, padding="max_length")
    with torch.no_grad():
        outputs = model(**inputs)
    pred = torch.argmax(outputs.logits, dim=-1).item()
    return pred  # Map to label using label_encoder.classes_

Limitations & Bias

• The model is trained on legal texts from specific municipalities and may not generalize to other domains or languages.
• Only entity types present in the training data are supported.
• The model expects entities to be marked with [E] ... [/E] in the input.

Citation

If you use this model, please cite:

@misc{legal-bert-classifier-refinement-abb,
  author = {S. Vercoutere},
  title = {Legal-BERT Entity Refinement},
  year = {2026},
  howpublished = {\url{https://huggingface.co/svercoutere/legal-bert-base-classifier-refinement-abb}}
}