# Pipeline for indonesia language

**URL:** <https://forum.rasa.com/t/pipeline-for-indonesia-language/44237>\
**Category:** Rasa Open Source\
**Created:** [June 3, 2021, 7:03am UTC](https://forum.rasa.com/t/pipeline-for-indonesia-language/44237 "2021-06-03T07:03:13Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![diazardian](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/diazardian/32/15200_2.png) [@diazardian](https://forum.rasa.com/u/diazardian)\
**Post date:** [June 3, 2021, 7:03am UTC](https://forum.rasa.com/t/pipeline-for-indonesia-language/44237/1 "2021-06-03T07:03:13Z")

</div>

What is the best pipeline settings for indonesian language?

---

<div class="post-metadata">

**Author:** ![koaning](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/koaning/32/6626_2.png) [@koaning](https://forum.rasa.com/u/koaning)\
**Post date:** [June 10, 2021, 11:23am UTC](https://forum.rasa.com/t/pipeline-for-indonesia-language/44237/2 "2021-06-10T11:23:20Z")

</div>

Hi Diaz,

it’s hard to upfront say “this pipeline always works best in language X” because it mainly depends more on the specific dataset that you have. A Dutch/English/German/Indonesian chatbot for customer service might need a different pipeline if it used for customer service compared to when it is used for HR.

That said, despite not being very familiar with Indonesian, I am trying to make more compatible tools for Rasa to ensure that we also support Non-English languages. Have you seen the [rasa nlu examples](https://github.com/RasaHQ/rasa-nlu-examples) project? It offers dense embeddings for Indonesian via the [BytePair Embeddings](https://rasahq.github.io/rasa-nlu-examples/docs/featurizer/bytepair/) and [FastText](https://rasahq.github.io/rasa-nlu-examples/docs/featurizer/fasttext/). Typically these embeddings boost entity detection performance.