Skip to content

Roadmap

BASA is building toward a modern NLP ecosystem for Indonesian and regional languages.

Our philosophy is to ship stable, well-tested features incrementally rather than pursuing rapid expansion without maintenance guarantees.

v0.1 — Text Normalization Foundation

Status: Stable Foundation

Core features:

  • quick() zero-configuration API
  • normalize() configurable normalization pipeline
  • Indonesian slang normalization
  • Repeated-character reduction (gkkkk → gk)
  • Conservative typo correction (opt-in)
  • Batch processing support
  • Full documentation with MkDocs
  • Automated CI and testing pipeline

v0.2 — Translation Utilities

Planned features:

  • Javanese → Indonesian translation
  • Indonesian → Javanese translation
  • Sundanese → Indonesian translation
  • Indonesian → Sundanese translation
  • Regional language dictionaries
  • Rule-based baseline translators

Goal:

Provide lightweight utilities for low-resource Indonesian regional language processing.

v0.3 — Indonesian NLP Evaluation Tools

Planned features:

  • Factual consistency evaluation for Indonesian summarization
  • ROUGE utilities for Indonesian datasets
  • BERTScore integration
  • IndoBART evaluation helpers
  • Explainable evaluation reports

Goal:

Make Indonesian NLP evaluation more accessible and reproducible.

v0.4 — Dataset Utilities

Planned features:

  • Dataset loaders
  • Dataset validators
  • Train/validation/test split helpers
  • IndoBERT preprocessing utilities
  • IndoBART preprocessing pipelines

Goal:

Reduce boilerplate for Indonesian NLP experiments and research projects.

v0.5 — Low-Resource Language Support

Planned features:

  • Synthetic data generation tools
  • Augmentation pipelines
  • Dictionary expansion utilities
  • Data quality checks
  • Support for additional Indonesian regional languages

Goal:

Lower the barrier to building NLP systems for underrepresented local languages.

Long-Term Vision

BASA aims to become a practical NLP toolkit for:

  • Indonesian text normalization
  • Regional language processing
  • Evaluation and benchmarking
  • Dataset preparation
  • Low-resource language development

The project prioritizes:

  • Developer experience
  • Conservative defaults
  • Production-friendly APIs
  • Strong documentation
  • Incremental and maintainable growth