Authors: Chenxi Liu, Yongqiang Chen, Tongliang Liu, Mingming Gong, James Cheng, Bo Han, Kun Zhang

Published on: February 06, 2024

Impact Score: 8.22

Arxiv code: Arxiv:2402.03941

Summary

• What is new: The introduction of COAT, which leverages large language models to assist in discovering high-level hidden variables from raw data for causal discovery.
• Why this is important: Traditional causal discovery approaches depend on high-quality variables given by experts, which are not always available in real-world applications.
• What the research proposes: COAT integrates large language models for extracting potential causal factors from unstructured data and uses these factors along with a causal learning module to discover causal relations.
• Results: COAT demonstrated its effectiveness through case studies on review rating analysis and neuropathic diagnosis.

Technical Details

Technological frameworks used: COAT (Causal representatiOn AssistanT)

Models used: Large Language Models, FCI algorithm

Data used: Unstructured raw data from real-world applications

Potential Impact

Data analytics, healthcare diagnostics, customer sentiment analysis, AI development companies

Want to implement this idea in a business?

We have generated a startup concept here: CausalNet.