Staying updated with relevant literature can be both time-consuming and challenging, often leaving researchers searching for efficient ways to gather and synthesize information. Artificial Intelligence (AI) is a transformative tool, that can increase both speed and accuracy of the literature searching process. This page introduces practical methods to use AI to find scholarly literature. It focuses on how generative AI can be used to design a search strategy, find references and map literature. By doing this, AI tools can be integrated into the research workflows. Whilst AI should be seen as positive development, its use should be consistent with the University’s AI Research Policy.
This page is structured by functions AI can assist with, suggesting tools and their help with the literature searching process.
Creating a Search Strategy
Tools: Microsoft Copilot
Generative AI is a great way to find keywords for a literature search. This can help widen your search, and act as a starting point for the construction of an advanced search string. It’s important, however, that you evaluate each term individually and ensure its relevant before including it within your search.
When writing your prompt, be as precise as possible and include as much context as possible The below example provides some key terms in response to: “find some alternative keywords for 'phonics' that can be used in a systematised literature review”
Creating A Search String
Once you have generated your keywords, Copilot can combine them using Boolean operators to create a search string suitable to use in academic databases. It crucial to evaluate the output to ensure its accuracy. Below is an example of a search string that addresses the prompt: "I am writing a search strategy to use in an academic database. I have three concepts. Here are the keywords for each of these concepts. Concept 1: health inequality, health inequalities, health disparity, health disparities. Concept 2: autism, autistic, autism spectrum disorder, ASD. Concept 3: health passport, autism health passport. Can you combine these keywords using the Boolean operators 'and' and 'or'?"
Alternatively, you can give Copilot an example of a search strategy and ask it to apply this to your research question. For example, you could input the following prompt: To find information to answer the question ‘How can health passports help reduce health inequalities for people with autism?’, the following search strategy was entered into an academic database: (health inequality or health inequalities or health disparity or health disparities) and (autism or autistic or ‘autism spectrum disorder’ or ASD) and (health passport or autism health passport). Please construct a search strategy to be used in an academic database which will find information to answer this question: 'How do climate change policies impact socio-economic inequalities in urban areas?'.
The strategy generated by Copilot can be used as a starting point for your database searches. It is essential to critically evaluate and refine the strategy by incorporating additional keywords, modifying existing ones, and employing search techniques such as phrase searching or truncation.
Finding References:
Tools: Scispace, Scite.ai, Elicit
AI can also be used to find references for literature reviews. Here, emphasis is not on generative content, but AI tools that provide effective information retrieval. Some generative AI tools may fabricate references or provide information that is not substantiated. Given this, always use tools designed for retrieving peer reviewed content. Many of these tools also generate short summaries of each source; these can be great for identifying relevant literature, but it is always advisable to consult the original source before citing it.
Scispace: This platform simplifies the search process by allowing researchers to quickly access academic papers and conference proceedings. It is free to use, but it is advisable to create an account. Its user-friendly interface supports advanced search queries and generates citation formats instantly. It is especially helpful for managing references and provides helpful summaries of each source. It is also responsive to free hand sentences.
Elicit: The platform offers a free basic plan and a paid plan with additional features. You need to create an account to access it. Elicit can help with various research tasks including locating relevant articles and summarising information. It searches for articles from Semantic Scholar which includes over 126 million academic papers (both open-access and non-open-access) sourced from various disciplines. Elicit also includes preprints from repositories such as arXiv and bioRxiv. The platform regularly updates its database to ensure researchers have access to the latest available papers.
Literature Mapping
Tools: Connected Papers, ResearchRabbit
Connected Papers is a powerful tool that helps researchers explore academic literature through visualisation. By entering a paper identifier (like a DOI or title), users can generate a graph that maps out related papers based on their similarity. This allows researchers to quickly identify key works, trends, and connections within a specific field. It includes:
Visual Graphs: Papers are displayed as nodes in a graph, with proximity indicating similarity, making it easy to spot clusters of related research.
Prior and Derivative Works: Users can explore foundational papers (prior works) and recent advancements (derivative works) connected to their topic.
Interactive Exploration: Clicking on a node reveals detailed information, including its abstract and citation data.
ResearchRabbit: The platform is free to use (with the creation of an account) and allows user to identify papers, track citations, and visualise scholarly networks. By using "seed papers," it initiates searches and explores connections between sources. It allows users to create maps of related literature, making it easier to understand how sources are interconnected. ResearchRabbit also facilitates collaboration by enabling users to share collections and leave comments.
Most academic publishers have AI policies designed to preserve academic integrity, transparency, and trust within the publishing process. Policies ensure AI tools, such as generative AI, protect ethical standards and credibility and help prevent plagiarism, misrepresentation, or inaccuracies that may arise from unregulated AI usage. Most policies also encourage the disclosure of AI, leading to a transparency that allows readers to understand the extent of AI's contribution.
AI policies of commonly used publishers are listed below:
