Optimization of Causal Discovery via LLMs

I am searching for an expert to amplify the efficacy of identifying causal relationships. The goal is to use Large language Models (LLMs) and test metamorphic techniques to improve causal discovery.

Key Tasks Include:

- Background knowledge: causal discovery, metamorphic testing, prompt engineering
Select a dataset from an existing benchmark
Design prompts (template) to generate causal relations/graphs
Define metamorphic relations for causal relations/graphs
Experiments (manually):
Without prompt engineering ask LLM to generate causal relations/graphs
With prompt engineering ask LLM to generate causal relations/graphs
With metamorphic testing ask LLM to generate causal relations/graphs
With prompt engineering + metamorphic testing ask LLM to generate relations/graphs
Compare results from all the above
Automate the experiments (software framework) – prompt-tuning framework for causal graph