YOUR GO-TO INTELLIGENCE RESOURCE FOR THE SCHOLARLY PUBLISHING INDUSTRY

EBSCO Information Services (EBSCO) is launching artificial intelligence (AI) beta programs to partner with customers and receive user input on the company’s AI efforts. The first beta test concluded in May 2024, and a second beta will begin in July 2024. The betas are intended to create efficiencies for researchers working within EBSCO Discovery Service^™ (EDS) and EBSCOhost^®. Beta testers are exploring two features — AI Insights, which will generate a short list of insights from full-text articles, and Natural Language Search, which allows researchers to search in their natural language, with user queries honored as questions in search.

EBSCO’s AI Insights feature generates on-demand, high-level summary statements, or ‘insights’ from full-text articles so researchers can quickly assess the relevancy of an article for their research topic. The insights are generated using a Retrieval Augmented Generation (RAG) method, where the AI is grounded on the full text to reduce hallucinations. Extensive variations of responses were verified by EBSCO librarians, customers, and end users via human-in-the-loop processes. During the beta testing period, customers were asked if the AI Insight was useful, and 90 percent of participants confirmed it was beneficial. EBSCO will be sharing the results of this beta in July 2024.

Natural Language Search, a beta release kicking off in July 2024, allows for searching in the researcher’s natural language and is intended to honor the users’ intent. The underlying search technology leverages EBSCO’s relevance ranking, and subject query expansion via the Unified Subject Index of linked data-controlled vocabularies and authoritative metadata, with the addition of AI query expansion for common words.

In Fall 2023, EBSCO announced upcoming plans to incorporate AI into the company’s products as well as guiding principles for responsible use of AI. In 2025, EBSCO will release two additional beta programs, AI Reference Assistance and Literature Review. To learn more about beta opportunities in the future, visit EBSCO Artificial Intelligence.