OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Data & Analytics

Bunkatopics

BunkaTopics is an MIT-licensed Python package for cleaning unstructured text, modeling and visualizing topics, and analyzing how text is framed. It helps developers explore corpora through topic descriptions and visual summaries, and supports filtering text for downstream workflows such as LLM fine-tuning.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Bunkatopics — OpenViST project profile in AI Data & Analytics.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use BunkaTopics when you have textual data to clean, group into topics or explore visually, especially when you want to choose the embedding, projection and clustering models in a Python workflow.

Who it is for

  • Developers building text-analysis workflows.
  • Analysts exploring themes in collections of articles or other documents.
  • Teams preparing and filtering text datasets before LLM fine-tuning.

Documented uses

  • Clean and filter text before downstream analysis or fine-tuning.
  • Discover and describe topics in a textual corpus.
  • Visualize computed topics to inspect a collection at a glance.
  • Explore a corpus through semantically customized framing axes.
Before you choose

Requirements & limits

The package is open source under the MIT license. The README describes installation with pip or from a Git clone.

  • The README identifies Python 3.9 and illustrates a Python package workflow.
  • Topic results depend on selected models; the documented example uses Sentence Transformers, UMAP and KMeans.
  • The embedding model should be appropriate for the text language, and model size is a consideration.
  • The optional LLM topic-naming example uses an OpenAI API key and recommends an instruction-tuned model.
  • Colab notebooks linked from the repository cover topic visualization, fine-tuning data cleaning and frame analysis.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Fit and visualize topics from a text corpus

Load text documents, fit BunkaTopics with selected models, retrieve topic descriptions and create a topic visualization.

Before you start

  • Python 3.9 is identified by the project badge.
  • Install the package with pip or from a Git clone.
  • Provide a list of text documents.
  • Select an embedding model and a projection model; the example uses Sentence Transformers and UMAP.
  • The example uses KMeans for clustering.

Steps

  1. Install BunkaTopics with pip install bunkatopics, or clone the repository and install it with pip install -e .

  2. Load or prepare the corpus as a list of text strings; the README example loads Medium article titles with the datasets package.

  3. Load a Sentence Transformers embedding model and configure UMAP with two components.

  4. Create Bunka with the embedding and projection models, fit it to the documents, then call get_topics with a clustering model such as KMeans.

  5. Call visualize_topics to render the computed topics.

Expected outcome

A set of topic descriptions and a visualization of topics computed from the text corpus.

Guide limitations

  • The selected embedding model should suit the language of the text; the README notes that language is detected automatically but says to use an adapted embedding model.
  • The README cautions that embedding model size should be considered.
  • The example specifies model choices and parameters; it does not prescribe one setup for every corpus.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
15 October 2026
Original website title
charlesdedampierre/BunkaTopics · Official homepage title · retained in its source language
How we review sourcesSuggest a correction