Scopus AI-Enabled Features (Scopus AI, Semantic Search, AI Query Builder)
Last updated on July 09, 2026
Scopus offers AI-enabled features that help users search, explore, and understand research content more efficiently. This page explains Scopus AI, AI Query Builder, and Semantic Search, including how each feature works, access and how to use it.
Scopus AI is an AI-driven research tool that uses the Scopus peer-reviewed research repository to help users understand and navigate unfamiliar academic content. Scopus AI generates summaries based on Scopus abstracts with references to help decipher complex content, facilitate deeper exploration, and provide academic insights.
Scopus AI is only available to users if it's part of their institutional subscription.
Scopus AI processes a query with an advanced sentence-level transformer, then locates relevant documents (published in Scopus since 2003) in our vector search index to identify the top pertinent documents. The identified documents are processed to generate the summary and other generative-AI features. The Scopus large language model (LLM) uses prompt engineering to ensure summaries are rooted in Scopus content and include references to maintain transparency.
What technologies does Scopus AI use?
Scopus AI is built using a combination of technologies from Elsevier and third-party providers. We use our own LLM technologies along with other LLMs, including the GPT developed by OpenAI for its ChatGPT tool. Our use of OpenAI GPT is private, so there is no data exchange or use of our data to train ChatGPT.
Scopus AI uses a vector search engine. It is powered by fine-tuned mini-language models that look at the words and sentences in a query and assign them a numerical value based on a language vector space. These vectors are compared to the abstracts of Scopus documents (published since 2003).
Scopus AI is hosted on Microsoft Azure. The frontend of the tool is built with a mix of JavaScript and CSS, while Python, Java, Elasticsearch and Langchain are used in the backend. To generate the list of Foundational papers, we use knowledge graph technology.
RAG Fusion leverages the intelligence of LLMs to expand upon an original query to explore new perspectives.
How do you decide which references to show in the Summary?
When entering a query in Scopus AI, the vector search engine uses a cosine similarity calculation to find the most relevant abstracts of documents (published since 2003). The prompt engineering uses factors such as relevancy, recency, and citation count to prioritize which sources are included in the summaries. The initial Summary may cite up to 20 papers, while the Expanded summary & Deep Research may considers up to 40.
How do you avoid confirmation bias and ensure Summaries represent a wide range of views?
Scopus AI is designed to provide a response based on a consensus from the academic literature. Prompt engineering uses factors such as relevancy, recency, and citation count to prioritize sources. Features like the Expanded summary, which uses RAG Fusion, may help reduce confirmation bias by exploring a query from different and alternative perspectives.
- Transparency: Scopus AI grounds its claims on the curated Scopus database and provides references.
- Quality: Scopus AI undergoes periodic evaluations using a quality framework.
- Reliability: Prompts help reduce outputs that are factually inaccurate or unreliable.
Why do some users have access to Scopus AI without a subscription?
Randomized user testing is one way that Scopus collects user feedback. Some users were randomly selected for access to Scopus AI. A user cannot request to be included in user testing because randomization helps ensure statistical validity.
Scopus AI is only available if it's part of your institutional subscription. Contact the administrator of your organization for information about Scopus AI subscriptions.
While Scopus AI strives to ground its summaries and generative AI features in trusted Scopus content, there may be occasional discrepancies. Users should not solely rely on Scopus AI outputs without conducting independent research. Do not enter personal, confidential, or sensitive information into Scopus AI.
Scopus AI is only available if it's part of your institutional subscription. Contact the administrator of your organization for information about Scopus AI subscriptions.
- From the Scopus homepage, select the 'Scopus AI' tab.
- Enter your query in the field.
- Press 'Enter' or select the search icon.
- From the Summary page, you can use the following options:
View the summary, references, and foundational documents
When you run a query, Scopus AI synthesizes abstracts from relevant documents to generate a topic summary. From the summary, you can view and export the references, foundational documents, and other associated documents.
The Expanded summary provides deeper insights into your query. Select 'Show all references' to show all references in a side panel, view abstracts, or export references.
Select 'Concept map' to show a visual map of the search results and navigate complex relationships.
Select 'Topic experts' to view prominent researchers related to your query. Select 'Preview profile' to open an Author profile preview.
Select 'Emerging themes' to explore new research directions related to your query and based on relevant abstracts from the last 24 months.
- Consistent themes - Research areas with a steady publication rate.
- Rising themes - Research areas with a growing academic presence.
- Novel themes - Evolving areas with recent interest.
Select any of the Go deeper links to open a new Summary and explore relevant related queries.
Deep Research helps users explore a topic in more depth by supporting a more detailed review of relevant Scopus content. Users should review the supporting references and use Scopus AI output as a starting point for further research rather than as a final source of truth. Deep Research may considers up to 40 references. Deep Research reports can be Downloaded as PDFs.
How does Scopus AI ensure data privacy and security?
Scopus AI strictly adheres to GDPR regulations to guarantee user privacy and avoids unnecessary data retention. No personal user information or chat history is retained on our systems unless compliantly done to enhance the product. We uphold RELX Responsible AI Principles, aiming to remove unfair bias, ensure accountability, and champion robust data governance.
AI Query Builder allows users to enter natural language queries or questions and convert them into keyword Boolean search strings for Scopus document search. AI Query Builder is available to all users. If Scopus AI is not included in your institution’s subscription, AI Query Builder access is limited to a daily metered allowance of 5 searches per day.
How does AI Query Builder work?
AI Query Builder co-pilot sanitises the natural language input, then breaks down a question into concepts and keywords. Concepts are combined with AND, while related keywords are combined with OR to construct a Boolean query. The generated query can be edited manually before running the document search.
Search results are generated from the Boolean query and are reproducible, which means you can save the query, set alerts and save results to a list. Review the generated query before searching to make sure it reflects your research question.
- From the Scopus homepage Document Basic search, enable the 'AI Query Builder' toggle.
- Enter your natural language query or question in the free text box.
- Select 'Generate query'.
- Review the generated query and edit it manually, if needed.
- Select 'Search' to run the document search.
Semantic Search allows users to search Scopus using natural language queries or questions. It uses semantic vector search with LLM embeddings to retrieve the most relevant results. Semantic Search is available to all users. If Scopus AI is not included in your institution’s subscription, Semantic Search access is limited to a daily metered allowance of 5 searches per day.
How does Semantic Search work?
Semantic Search compares the meaning of the user’s query with document content using vector search and LLM embeddings. It retrieves the most relevant results and supports additional tools to help users refine, review, and understand the result set.
Semantic Search can retrieve top relevant results using natural language queries or questions.
Use a natural language query or question to retrieve semantically relevant results. From the results page, you can refine and review results using the following options:
Use filters to narrow the result set. Available filters include document type, subject area, language, publication year, and open access.
Use the action toolbar to work with selected results. Available actions include Export, Citation Overview, and Download (Bulk) PDF.
Semantic Search Co-pilot sanitises the query, interprets the query intent, allows multi-language search, supports implicit filters (Date, Country, Document Type)
If enabled, the AI Summary option allows users to generate an AI summary based on the default top 20 results. This helps users understand the result set by providing an AI-generated overview of relevant content. Review the supporting documents and references before using the summary in your research.
Did we answer your question?
Related answers
Recently viewed answers
Functionality disabled due to your cookie preferences