# milanlproc/contextualized-topic-models

**Attribution required: if you use, quote, or summarise this content, you must credit and link back to [awesome-repositories.com](https://awesome-repositories.com/repository/milanlproc-contextualized-topic-models).**

1,271 stars · 154 forks · Python · MIT

## Links

- GitHub: https://github.com/MilaNLProc/contextualized-topic-models
- awesome-repositories: https://awesome-repositories.com/repository/milanlproc-contextualized-topic-models.md

## Topics

`bert` `embeddings` `multilingual-models` `multilingual-topic-models` `neural-topic-models` `nlp` `nlp-library` `nlp-machine-learning` `text-as-data` `topic-coherence` `topic-modeling` `transformer`

## Description

Contextualized topic modeling is a framework that integrates deep learning architectures with statistical word frequency distributions to extract coherent themes from large document collections. By combining pre-trained transformer-based embeddings with variational inference, the system identifies hidden patterns in text while maintaining the interpretability of traditional generative models.

The library distinguishes itself by mapping diverse languages into a shared semantic space, enabling topic discovery and classification across multilingual datasets without requiring language-specific training data. It further supports supervised thematic modeling, allowing users to incorporate known category labels or human feedback during the training process to steer the latent space toward specific outcomes.

The toolkit provides a comprehensive suite of utilities for natural language data preparation, including text cleaning and the transformation of corpora into dual formats suitable for both dense vector representations and sparse frequency analysis. These capabilities support a range of approaches, including zero-shot discovery and iterative, human-in-the-loop classification workflows.

## Tags

### Artificial Intelligence & ML

- [Topic Modeling Libraries](https://awesome-repositories.com/f/artificial-intelligence-ml/topic-modeling-libraries.md) — Provides a framework that combines BERT embeddings with statistical topic models to extract themes from large document collections.
- [Contextual Embeddings](https://awesome-repositories.com/f/artificial-intelligence-ml/natural-language-processing/word-embeddings/contextual-embeddings.md) — Integrates pre-trained transformer-based contextual embeddings to capture semantic word relationships for topic modeling.
- [Variational Autoencoders](https://awesome-repositories.com/f/artificial-intelligence-ml/variational-autoencoders.md) — Uses variational autoencoder architectures to approximate the posterior distribution of latent topics from document representations.
- [Text Classification Training](https://awesome-repositories.com/f/artificial-intelligence-ml/custom-model-training/detection-model-training/convolutional-neural-network-training/text-classification-training.md) — Enables topic classification across multiple languages without requiring language-specific training data.
- [Topic Models](https://awesome-repositories.com/f/artificial-intelligence-ml/generative-ai-resources/diffusion-visual-models/generative-ai-models/latent-space-generative-models/topic-models.md) — Generates contextualized topics by merging pre-trained language representations with word frequency models. ([source](https://github.com/milanlproc/contextualized-topic-models#readme))
- [Latent Conditioning Mechanisms](https://awesome-repositories.com/f/artificial-intelligence-ml/latent-conditioning-mechanisms.md) — Constrains the latent space using known category labels during training to steer thematic outcomes.
- [Text Document Classification](https://awesome-repositories.com/f/artificial-intelligence-ml/machine-learning-classification/text-document-classification.md) — Supports interactive workflows where human feedback refines and labels document clusters during classification. ([source](https://github.com/milanlproc/contextualized-topic-models#readme))
- [Multilingual Text Embeddings](https://awesome-repositories.com/f/artificial-intelligence-ml/multilingual-text-embeddings.md) — Leverages multilingual word embeddings to discover consistent topics across diverse linguistic datasets.
- [Latent Dirichlet Allocations](https://awesome-repositories.com/f/artificial-intelligence-ml/topic-modeling-libraries/latent-dirichlet-allocations.md) — Combines traditional generative probabilistic models with deep learning to improve the coherence of discovered thematic clusters.
- [Supervised Topic Modeling](https://awesome-repositories.com/f/artificial-intelligence-ml/topic-modeling-libraries/supervised-topic-modeling.md) — Incorporates known labels during training to guide the identification of themes toward specific categories.
- [Zero-Shot Topic Modeling](https://awesome-repositories.com/f/artificial-intelligence-ml/topic-modeling-libraries/zero-shot-topic-modeling.md) — Detects topics in unfamiliar datasets or new languages using multilingual embeddings without requiring specific training. ([source](https://github.com/milanlproc/contextualized-topic-models#readme))

### Data & Databases

- [Natural Language Preprocessing Utilities](https://awesome-repositories.com/f/data-databases/natural-language-preprocessing-utilities.md) — Cleaning raw text and transforming it into dual formats suitable for both modern vector-based models and traditional word frequency analysis.
- [Thematic Categorization](https://awesome-repositories.com/f/data-databases/text-processing-utilities/text-processing-tools/selective-text-processors/ai-text-transformers/thematic-categorization.md) — Identifies hidden patterns and coherent topics by combining contextual embeddings with statistical modeling.
- [Vector Space Models](https://awesome-repositories.com/f/data-databases/vector-storage/text-vectorizers/vector-space-models.md) — Maps diverse languages into a shared semantic vector space to enable cross-lingual topic discovery.