This guide focuses on two categories of AI-generated research tools: citation-based literature mapping tools and AI-powered synthesis tools for literature reviews. Citation mapping tools help researchers discover and visualize related academic literature. AI synthesis tools assist in finding, summarizing, and analyzing relevant research by extracting key insights, identifying consensus, and organizing evidence across multiple studies.
Academic deep research: AI-powered research tools (such as Undermind, Elicit, Consensus) that use deep search specifically on scholarly databases to produce orientation artifacts—reports with features like timeline visualizations, foundational papers, research gaps, and evidence tables—designed to help researchers quickly understand a new field rather than generate copy-paste literature reviews.
Artificial intelligence (AI): technology that enables computers and machines to simulate human intelligence and problem-solving capabilities (IBM).
Citation chasing: A search technique that follows citation networks by examining both backward citations (references cited by a paper) and forward citations (papers that cite a paper) to discover related scholarly works, often used as part of deep search workflows.
Copilot: used to understand and interpret the text written by experts, helping the platform to identify key points and summarize the information effectively. (not to be confused with Microsoft Copilot)
CORE: a free, open-access, and comprehensive collection of millions of scholarly articles aggregated from thousands of repositories and journals worldwide.
Crossref: A Digital Object Identifier (DOI) Registration Agency enabling persistent cross-platform citation linking.
Deep search (also called agentic search): An iterative search methodology that performs multiple rounds of retrieval, blending keyword search, semantic/vector search, and citation chasing, with AI-powered relevance judgments at each step. Unlike traditional one-shot searches, deep search automatically explores different query paths and follows citation trails to achieve higher recall and precision, though it requires more processing time.
Force directed graph: used to visualize the relationships between scholarly articles, where each node represents a paper and the edges represent citations or other connections between the papers. The force-directed layout algorithm simulates attractive and repulsive forces between the nodes, causing closely related papers to cluster together and less related papers to be pushed apart.
Generative AI: refers to a class of artificial intelligence (AI) technologies that produce outputs such as text, images, datasets, or other media in response to user prompts (Carnegie Council for Ethics in International Affairs, 2024).
Knowledge graph: structured database that represents academic information as a network of interconnected entities (e.g., papers, authors, concepts) and their relationships. This structure allows for more sophisticated querying and analysis of academic literature.
Large language model (LLM): a deep learning model trained on vast amounts of text data to understand and generate human language.
Machine learning: the development and application of algorithms that enable computer systems to automatically learn and improve from experience without being explicitly programmed, by identifying patterns and making predictions or decisions based on input data.
Multi-origin graphs: A feature in citation mapping tools like Connected Papers that allows users to iteratively add multiple seed papers to refine their literature search, with each addition regenerating the similarity graph to show papers relevant to all selected origins collectively.
OpenAlex: a free, open-source, and comprehensive index of scholarly works, research organizations, publication venues, and author information that aims to make scholarly data more accessible, interoperable, and reusable for the scientific community and beyond; it was launched in January 2022 as a successor to Microsoft Academic Graph (MAG) and sources its data from various open databases, including Crossref, PubMed, and ORCID, as well as web pages related to scholarly works.
Retrieval-augmented generation (RAG): a technique that allows AI research assistant tools to generate more accurate and contextual answers to questions. It works by using a large language model (LLM) – a deep learning model trained on vast amounts of text data to understand and generate human-like language – to comprehend the question. The LLM then retrieves relevant information from a knowledge base to help formulate a direct answer, rather than just providing a list of search results (Alan Zeichick, 2023).
SciScore: an automated tool that assesses the reproducibility and transparency of scientific research by analyzing the methods, materials, and reporting practices described in published articles.
Semantic Scholar: a free, AI-powered research tool that provides a comprehensive database of academic papers, utilizing machine learning techniques to extract key insights and identify connections between scholarly works. It leverages data from various sources, including partnerships with publishers, web crawling, and openly available databases like PubMed, Crossref, and ORCID.
Semantic search: a type of search that focuses on understanding the meaning and intent behind a user's query, rather than just matching keywords.
Vector search: a technique that represents data points, such as text documents or images, as high-dimensional numerical vectors and uses similarity measures to find the most relevant items in a large dataset.
Zotero: a free, open-source reference management software that helps researchers collect, organize, cite, and share their research sources.
