Abstract: In order to improve the efficiency and quality of indexing, an experimental method using generative artificial intelligence is applied to address the shortcomings of traditional rule-based and probabilistic indexing software. A scheme for indexing using the retrieval-augmented generation of generative AI is proposed. Specifically, the journal volume is used as a unit, and the full text is subject-indexed based on indexing words derived from the abstracts of the papers by utilizing a large language model. The paper presents the design of an article abstract indexing database system, which can realize textual concept extraction and subject indexing based on the large model of massive knowledge emergence and reasoning capabilities, as well as key information and emerging knowledge extracted from abstracts. Additionally, this paper explores practical ways to associate index and abstract with academic norms and evaluation, demonstrating the potential value of generative AI in the field of indexing. It provides insights into promoting the application of generative AI technology in library and information science.

Key words: Index, Subject index, Generative artificial intelligence, Retrieval-augmented generation, Academic norm, Academic evaluation