Chen is a cost-effective, fast, and easy-to-implement text classification algorithm.
It can build unsupervised text classifiers for various use cases with minimal data and computational requirements, solely through prompt engineering and pre-trained word embeddings. No heavy data curation or model training needed.
The usage is very simple, including the following steps:
- Construct vocabularies for specific use cases through prompt engineering.
- Convert these vocabularies into a classification kernel using an embedding model.
- Process and classify the text to be categorized.