# Bahasa-Indo-NLP-Dataset: A Curated Index, Not a Dataset, for Indonesian NLP > This repository is a hand-maintained list of Indonesian NLP libraries, corpora, and models. It is a starting point for researchers and developers, but it is not a dataset itself and carries no versioning or maintenance guarantees. **keyreply/Bahasa-Indo-NLP-Dataset** — Bahasa-Indo-NLP-Dataset - Repository: https://github.com/keyreply/Bahasa-Indo-NLP-Dataset - Stars: 100 · Forks: 28 - Language: Unknown - License: not declared - Published: 2026-08-17 · Updated: 2026-08-18 · Language: en - Canonical page: https://hysenlabs.com/projects/keyreply-bahasa-indo-nlp-dataset ## What This Repository Actually Is Despite the name, this is not a dataset. It is a curated list, a hand-written table of links to Indonesian natural language processing resources. The README calls itself a collection of libraries, dictionaries, and corpus, but the actual content is a set of markdown tables. Each row points to an external GitHub repository, a paper, or a downloadable file. The intended user is someone who needs a map of what exists for Bahasa Indonesia before choosing a tool. The repository has no code of its own, no data files, and no package to install. It is a directory, and that is its only function. ## The Scope and Structure of the Index The README organizes resources into nine categories: libraries, sentiment analysis, POS tagging, named entity recognition, stemmers, word embeddings, question answering, dictionaries and parallel corpora, downloadable text corpora, WordNet, universal dependency treebanks, pre-trained word vectors, and a grammar resource framework. Each category is a markdown table with columns for name, description, language, license, and a link. For example, the stemmer section lists Sastrawi, a PHP library, and the NER section lists indonesia-ner, a Java project. The corpus section includes a 29.64 GB CoNLL 2017 annotated Indonesian corpus from LINDAT/CLARIN. The index is broad but shallow: it lists resources without rating them or comparing their quality. ## How to Use It: Commands and Config Keys Are Absent There is no installation command, no configuration file, and no API to call. Using this repository means opening the README, finding a row that matches your task, and following the external link. For instance, if you need a stemmer, you click the Sastrawi link and then install that PHP package separately. If you need pre-trained vectors, the README gives direct URLs: a fastText binary and text file at s3-us-west-1.amazonaws.com/fasttext-vectors/wiki.id.zip and a text-only version at wiki.id.vec. For word2vec, there is a Google Drive link to a 402 MB file with 300 dimensions. The only practical action is copying those URLs into a browser or a download tool. The repository itself offers no commands to run. ## A Real Limitation: Incomplete Metadata and No Maintenance Signal Several entries lack license fields. The sentiment analysis library python-sentianalysis-id has an empty license column, as does the Open NLP POS tagger and the word embedding web app. That is a problem for production use, because you cannot legally reuse a library without knowing its license. The repository also has no release history and no last-push date in the metadata we have. The README says it is pre-alpha for the bahasa library and describes other resources as experimental. One entry, the Indonesian word embedding app, is a demonstration, not a library. The repository is a snapshot, and the snapshot may be years old. You cannot tell from the list which links are dead or which projects are abandoned. ## The Alternative: kmkurn/id-nlp-resource The README itself points to a competing list at https://github.com/kmkurn/id-nlp-resource. That repository is another curated index of Indonesian NLP resources. The difference is that kmkurn's list is more focused on academic papers and models, while this one mixes libraries, corpora, and services. The README explicitly says to look there if this list does not have what you need. For a developer, the practical choice is to check both lists and cross-reference. Neither list is a substitute for searching the web or checking the actual state of a linked project. ## Maintenance and Upgrade Cost: What You Should Verify The repository is not archived, but there are no recent releases and no push dates in the available metadata. The README says it always welcomes pull requests, which suggests it is open to contribution, but that is not the same as active maintenance. The cost of using this list is the time you spend verifying each link. Licenses vary: MALINDO_Morph is CC BY-NC-SA 4.0 TH, TALPCo is CC BY 4.0, the CoNLL corpus is CC BY-NC-SA 4.0 TH, and the UD treebank is CC BY-NC-SA 3.0 US. The non-commercial clauses in several licenses mean you cannot use those resources in a commercial product without separate permission. The fastText vectors are CC BY-SA 3.0, which has share-alike implications. Before adopting any resource, open its own repository and read its license file. ## A Point of View on the Trade-offs This repository is a useful bookmark, not a tool. Its strength is that it gathers many types of resources in one place, from stemmers to grammar frameworks. Its weakness is that it offers no quality signal and no freshness guarantee. A table with empty license cells is a warning sign, not a detail. If you are a researcher exploring the field, this list saves you an afternoon of searching. If you are an engineer choosing a dependency for a product, you will still need to evaluate each candidate on your own. The repository does not tell you which library works well, which is maintained, or which has a permissive license. It only tells you that the resource exists. ## Conclusion Adopt this repository if you are starting Indonesian NLP and need a single page of links to libraries, corpora, and models. Do not adopt it if you expect a downloadable dataset, active maintenance, or versioned releases. Before relying on any listed resource, verify its license and last update directly at the linked source, because several entries lack license fields and the repository itself has no release history. Use it as a directory, not as a dependency. ## FAQ ### What is an NLP dataset, according to this index? The index lists several kinds. Indonesian-annotated-conll17 is a 29.64 GB CoNLL Universal Dependency Parsing corpus with automatic segmentation, tokenization, and morphological and syntactic annotations, and ID-OpinionWords is a list of positive and negative opinion words for sentiment analysis. ### Is bahasa the same as Indonesian? The index treats the two as the same language. It lists a Python toolkit named bahasa for Bahasa Indonesia and a PHP stemmer described as a high quality stemmer library for Indonesian Language (Bahasa). ### How do you say I love you in Bahasa, and can this index help? Not for that. The repository is an index of tools and corpora rather than a phrasebook. The closest entry is MALINDO_Morph, a morphological dictionary for Malay and Indonesian with English-Malay and English-Indonesian pairs. ### What are the top 3 languages spoken in Indonesia? The repository does not discuss languages spoken in Indonesia. Its parallel resources cover English-Malay, English-Indonesian through MALINDO_Morph, and Japanese to Indonesian through TALPCo. ## Sources - [Official README](https://github.com/keyreply/Bahasa-Indo-NLP-Dataset#readme) - [Project repository](https://github.com/keyreply/Bahasa-Indo-NLP-Dataset) - [README](https://github.com/keyreply/Bahasa-Indo-NLP-Dataset/blob/master/README.md) - [Releases](https://github.com/keyreply/Bahasa-Indo-NLP-Dataset/releases) --- Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/keyreply-bahasa-indo-nlp-dataset