A Prompt Engineering Approach for Structured Data Extraction from Unstructured Text Using Conversational LLMs
Open Access
- 22 December 2023
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 183-189
Abstract
This paper aims to extract structured information from unstructured text written in natural language. This extracted information can then be stored in a database and queried using database access languages such as SQL to derive meaningful answers to questions that might arise around a specific activity related to that unstructured text. For example, an email chain discussing a planned trip could be used to extract a record with fields such as who is taking the trip, where they are departing from, when they are departing, where they are going, and when they are arriving. This information could then be used to create a relation of trips that could be queried by SQL. This paper uses a prompt engineering approach using conversational LLMs for extracting the relevant information related to travel and store the information into relational databases which can then be queried using SQL or any other query language. This initiative holds the promise to revolutionize the storage and retrieval of information with minimal effort. Currently, unstructured text is difficult to query and analyze. By extracting structured information from this text, it becomes much easier to work with.Keywords
This publication has 10 references indexed in Scilit:
- Extracting accurate materials data from research papers with conversational language models and prompt engineeringNature Communications, 2024
- Flexible, Model-Agnostic Method for Materials Data Extraction from Text Using General Purpose Language ModelsarXiv, 2024
- Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language ProcessingACM Computing Surveys, 2023
- Dataset of solution-based inorganic materials synthesis procedures extracted from the scientific literatureScientific Data, 2022
- ChemDataExtractor 2.0: Autopopulated Ontologies for Materials ScienceJournal of Chemical Information and Modeling, 2021
- Recipes for Building an Open-Domain Chatbot2021
- Data-driven materials research enabled by natural language processing and information extractionApplied Physics Reviews, 2020
- Machine Learning in Materials Discovery: Confirmed Predictions and Their Underlying ApproachesAnnual Review of Materials Research, 2020
- Biomedical named entity recognition using deep neural networks with contextual informationBMC Bioinformatics, 2019
- ChemDataExtractor: A Toolkit for Automated Extraction of Chemical Information from the Scientific LiteratureJournal of Chemical Information and Modeling, 2016