Cerberus is an app that reduces the annotation burden of linguists. It does this by making it easy for linguists to apply state-of-the-art natural language processing models to their data. Given some initial data, these models learn to perform linguistic annotation tasks themselves. They can then automatically perform those tasks on a much larger dataset, reducing the manual labour of a linguist. The models are not perfect and are designed to help bootstrap a linguistic project.
Cerberus currently supports the following tasks:
- POS tagging: Assigning a syntactic category to each word.
- Translation: Automatically translating from one language to another.
- Classification: Assigning a user-defined label to a word, sentence or paragraph.
Coming soon:
- Spelling correction: Correcting misspelt words.
- Morphological analysis: Assigning morphosyntactic features to each word.
- Language modeling: Generating grammatical sentences.
Cerberus is built on top of AllenNLP and Streamlit.
