[2402.01454] Integrating Large Language Models in Causal Discovery: A Statistical Causal Approach

PDF view of the paper entitled Merging the Long Language Models in the Causive Discovery: Statistical Causative Approach, by Masayuki Takayama and 6 other authors

PDF HTML (experimental) view

a summary:In the practical statistical discovery (SCD), the integration of expert knowledge in the field as restrictions in the algorithm is important for reasonable causal models that reflect the extensive knowledge of the field experts, despite the challenges in the systematic acquisition of basic knowledge. To overcome these challenges, this paper suggests a new way of causal reasoning, as SCD is manufactured and causal inference is KBCI with a large language model (LLM) through the statistical “SCP) to increase the LLMS and increase the previous knowledge of SCD. Experiences in this work revealed that the results of LLM-KBCI and SCD are reinforced with LLM-KBCI approaches the ground facts, more than SCD without prior knowledge. The future of this proposed method through important areas such as health care, we also discuss accurate restrictions, the risks of critical errors, the expected improvement of technologies about LLMS, and the realistic integration of experts ’tests in this automatic process, with SCP simulation under different circumstances in the scenarios of success and failure. Challenges such as data collection and restrictions, which shows LLMS capabilities to improve data based on data -based on various scientific fields.

The symbol used in this work is available to the public on: URL http