Design Philosophy

BASA is built around a simple principle:

Conservative defaults, developer-friendly APIs, and practical tools for Indonesian NLP.

The project prioritizes reliability and maintainability over rapid feature expansion.

Conservative by Default

Text normalization can be destructive.

Proper nouns, technical terms, abbreviations, and domain-specific vocabulary should not be modified without explicit user intent.

For this reason:

normalize(
    text,
    apply_typo=False
)

Typo correction is always opt-in.

Most users should not need to read extensive documentation before becoming productive.

The simplest workflow should look like this:

from basa import quick

quick("gw gk ngerti bngtttt")

BASA aims to provide sensible defaults while still allowing advanced configuration when necessary.

BASA focuses primarily on Indonesian language processing.

Many global NLP tools are designed for English-first workflows, leaving Indonesian-specific problems underrepresented.

Examples include:

The project addresses these problems directly.

Indonesia has hundreds of local languages, many of which remain underrepresented in modern NLP ecosystems.

Future BASA releases aim to support:

The long-term goal is practical tooling rather than academic benchmarks alone.

BASA prefers explicit configuration over hidden behavior.

For example:

normalize(
    text,
    apply_typo=True,
    lowercase=False
)

Users should always understand which transformations are being applied to their data.

Breaking changes are introduced cautiously.

The public API is intentionally small:

from basa import (
    normalize,
    quick,
    slang,
    typo,
)

A small API surface is easier to maintain, document, and support over time.

BASA follows an incremental development model.

Instead of shipping many unfinished features, the project focuses on:

The objective is not to become the largest NLP framework, but to become a dependable toolkit for Indonesian and regional language processing.