Skip to content

Quick Start

BASA provides two main entry points:

  • quick() for zero-configuration usage
  • normalize() for fine-grained control

Quick Normalization

The simplest way to clean Indonesian informal text:

from basa import quick

quick("gw gk ngerti bngtttt sihhhh!!!")

Output:

'saya tidak mengerti banget sih!'

Fine-Grained Control

Use normalize() when you need more control over the pipeline:

from basa import normalize

normalize(
    "GW GKKKK NGERTIII BNGTTTT!!!!!",
    apply_slang=True,
    apply_typo=False,
    lowercase=True,
    normalize_punctuation=True,
)

Output:

'saya tidak mengerti banget!'

Batch Processing

BASA supports lists of strings out of the box:

texts = [
    "gw gk ngerti",
    "udh makan blm?"
]

normalize(texts)

Output:

[
    'saya tidak mengerti',
    'sudah makan belum?'
]

Preserving Original Casing

Disable automatic lowercasing when case information matters:

normalize(
    "Jokowi pergi ke Jakarta",
    lowercase=False
)

Output:

'Jokowi pergi ke Jakarta'

Typo Correction

Typo correction is disabled by default.

Load your vocabulary first:

from basa import normalize, typo

typo.add_to_vocab({
    "makan",
    "minum"
})

Then enable typo correction:

normalize(
    "gw mkan",
    apply_typo=True
)

Output:

'saya makan'

Why Is Typo Correction Opt-In?

Typo correction can accidentally modify:

  • technical terms
  • proper nouns
  • abbreviations
  • domain-specific vocabulary

To prevent destructive transformations, BASA keeps typo correction disabled by default.

Next Steps

  • Read the API Reference for detailed documentation.
  • Explore the Guides section for practical NLP workflows.
  • Check the Roadmap to see upcoming features.