ChatGPT Chemistry Assistant for Text Mining and the Prediction of MOF Synthesis

We use prompt engineering to guide ChatGPT in the automationoftext mining of metal-organic framework (MOF) synthesis conditionsfrom diverse formats and styles of the scientific literature. Thiseffectively mitigates ChatGPT's tendency to hallucinate information,an issue that previously made the use of large language models (LLMs)in scientific fields challenging. Our approach involves the developmentof a workflow implementing three different processes for text mining,programmed by ChatGPT itself. All of them enable parsing, searching,filtering, classification, summarization, and data unification withdifferent trade-offs among labor, speed, and accuracy. We deploy thissystem to extract 26 257 distinct synthesis parameters pertainingto approximately 800 MOFs sourced from peer-reviewed research articles.This process incorporates our ChemPrompt Engineering strategy to instructChatGPT in text mining, resulting in impressive precision, recall,and F1 scores of 90-99%. Furthermore, with the data set builtby text mining, we constructed a machine-learning model with over87% accuracy in predicting MOF experimental crystallization outcomesand preliminarily identifying important factors in MOF crystallization.We also developed a reliable data-grounded MOF chatbot to answer questionsabout chemical reactions and synthesis procedures. Given that theprocess of using ChatGPT reliably mines and tabulates diverse MOFsynthesis information in a unified format while using only narrativelanguage requiring no coding expertise, we anticipate that our ChatGPTChemistry Assistant will be very useful across various other chemistrysubdisciplines.

Automated summary

We use prompt engineering to guide ChatGPT in automating the text mining of metal-organic framework (MOF) synthesis conditions from scientific literature. This mitigates ChatGPT's tendency to hallucinate information and allows for the extraction of synthesis parameters and the creation of a machine-learning model for predicting experimental outcomes. The ChatGPT Chemistry Assistant, developed through this process, is expected to be highly useful in various chemistry subdisciplines.