Indonesian BERT Named Entity Recognition (NER) Pipeline

This repository contains a fine-tuned IndoBERT model for Indonesian Named Entity Recognition, wrapped in a HuggingFace-compatible pipeline for easy usage.

🚀 Quick Start

from transformers import pipeline

ner = pipeline(
    "ner",
    model="smapproyek/indobert_ner_p1",
    tokenizer="smapproyek/indobert_ner_p1",
    aggregation_strategy="simple"
)

text = "Hadiri KTT G20, Momen Gibran Bersama Erdogan hingga PM Jepang"
entities = ner(text)

print(entities)
# Output: [{'entity_group': 'EVENT', 'score': np.float32(0.9938912), 'word': 'ktt g20', 'start': 7, 'end': 14}, {'entity_group': 'PERSON', 'score': np.float32(0.80416846), 'word': 'gibran', 'start': 22, 'end': 28}, {'entity_group': 'PERSON', 'score': np.float32(0.99820226), 'word': 'erdogan', 'start': 37, 'end': 44}, {'entity_group': 'REPRESENTATIVE_PERSON', 'score': np.float32(0.9186572), 'word': 'pm jepang', 'start': 52, 'end': 61}]

🎯 Supported Entity Types

• PERSON — Names of individual humans
• REPRESENTATIVE_PERSON — Government/public roles
• POLITICAL_ORGANIZATION — Constitutional political institutions
• ORGANIZATION — Companies, universities, BUMN , NGOs
• LAW — UU, UUD, Kepres
• EVENT — National events, gov programs