TaxonGPT: Taxonomic Classification Using Generative Artificial Intelligence
Preprint
Abstract
To address the challenges of consistency and quality in taxonomic classifications, we have developed a Python program called TaxonGPT, that utilizes the natural language processing capabilities of generative artificial intelligence (gAI, specifically, ChatGPT-4o) to generate taxonomic descriptions and taxonomic keys. To counter the propensity for large language model gAIs to “hallucinate”, we use knowledge graph semantic representation and an error-checking module to ensure that accurate taxonomic descriptions and keys are obtained. In this paper, we describe how TaxonGPT embeds ChatGPT-4o’s responses as outputs. We also report on benchmark tests for accuracy, efficiency and reproducibility. These tests demonstrate that TaxonGPT excels in generating taxonomic keys and descriptions.Keywords
This publication has 39 references indexed in Scilit:
- Can ChatGPT Replace Traditional KBQA Models? An In-Depth Analysis of the Question Answering Performance of the GPT LLM Family2023
- Generative artificial intelligence and its applications in materials science: Current situation and future perspectivesJournal of Materiomics, 2023
- Science in the age of large language modelsNature Reviews Physics, 2023
- Still not enough taxonomists: reply to Joppa et al.Trends in Ecology & Evolution, 2012
- A Metadata Best Practice for a Scientific Data RepositoryJournal of Library Metadata, 2009
- Taxonomy in a Changing World: Seeking Solutions for a Science in CrisisSystematic Biology, 2007
- The Taxonomic Impediment Overcome: NSF's Partnerships for Enhancing Expertise in Taxonomy (PEET) as a ModelSystematic Biology, 2003
- A comparison of two methods of character constructionCladistics, 1995
- On character coding for phylogeny reconstructionCladistics, 1995
- The measurement of diversity in different types of biological collectionsJournal of Theoretical Biology, 1966