This application makes it easier to annotate natural language text with liguistic devices discussed in Jeanne Fahnestock's Rhetorical Style - Uses of language in persuasion.
It leverages ChatGPT as a model in the loop to aid the human annotator in the annotation process.
It takes as input constituency parse trees, and outputs a JSON tree with annotations for each sentence.
- MySQL 5.6.40 or higher
- Python 3/Flask
- ChatGPT API key
- Javascript/ES6
This folder contains all the code used in analysis and prompting GPT.
The features and properties were extracted from Rhetorical Style: The use of language in persuasion, Fahnestock (2011).
The json files containing all terms, definitions, and examples are in:
- APP/features-sentence.json
- APP/features-word.json