The article showcases the application of Artificial Intelligence (AI) in the realm of literature review through detailed tool demonstrations. Its primary aim is to introduce practical tools and clarify operational procedures to support teachers and students in enhancing research efficiency and broadening academic perspectives through the use of AI technology. The text systematically categorizes the AI-assisted literature search tools based on their distinct characteristics and scopes of application, providing comprehensive introductions to their functionalities and common usage of different tools. Additionally, it explores the academic potential of AI application tools in terms of improving efficiency and facilitating interdisciplinary research.

 

Keywords: Artificial Intelligence, Literature review, Tools

 

 

With the advancement of society, academic topics are becoming increasingly complex, and the integration of multidisciplinary knowledge has become a new trend. We encourage both teachers and students to transcend the confines of a single field in their academic research, instead integrating knowledge from multiple domains and disciplines to foster innovation. To this end, this article introduces several practical AI tools for literature review, aiming to assist researchers in exploring new knowledge in various fields more efficiently.

 

 

Information overload and the exponential growth of literature pose significant challenges to researchers, particularly inefficiently screening vast amounts of information to identify high-quality knowledge within a limited time. Furthermore, the rapid updating of new knowledge and the trend towards interdisciplinary and transdisciplinary evolution in academic research require researchers to constantly maintain the ability to learn new knowledge. In this context, the following text will outline how to efficiently utilize AI tools to assist literature reviews.

 

 

AI literature review tools, based on their algorithms and drawing from paper databases as information sources, provide services such as dynamic reviews and personalized literature recommendations. These tools visualize the results for users to clarify the associations between articles, offering insights and directions for subsequent knowledge exploration.

 

This article categorizes common AI literature review tools into three types based on their unique approaches and advantages: 1. Tools for literature-driven exploration of new fields, such as Research Rabbit and Connected Papers; 2. Tools for query-based literature retrieval, such as Elicit, and Consensus; 3. Tools for enhancing efficiency in literature review, such as Explainpaper. The following section will demonstrate the usage of these tools in different scenarios.

 

 

 

ResearchRabbit, based on the citation literature of uploaded articles, can visualize the association network between research interests and similar articles or authors through graphical representation.

 

Step 1

Import the articles you are interested in. After clicking "Add Papers," you can initiate your research by searching for a starting point using the title, DOI, keywords, or other methods.

 

Step 2

 

In the left-hand sidebar, two primary options offer recommendations tailored to your requirements. You may choose to “Explore Papers” or “Explore People”.

 

 

Step 3

 

The sharing feature facilitates the establishment of a project group, allowing added members to collaboratively edit the paper library.

 

 

 

The advantage of Connected papers lies in its algorithm-based visualization of literature similarity, rather than relying on citation relationships,  thereby improving search efficiency.

 

Step 1

 

Select a paper within your academic field of interest and click to generate a visual graph. The tool will recommend literature based on similarity, with each node representing a paper. The size of the node indicates the number of citations, and the nodes are arranged based on the degree of similarity.

 

Step 2

 

The tool provides two key features: “Prior Works” and “Derivative Works”. The “prior Works” feature is useful for identifying pioneering works in the field, while “Derivative Works” is employed for searching recently published relevant literature in the domain.

 

 

 

 

Elicit can recommend literature based on user inquiries. It assists researchers in efficiently screening and extracting details such as literature summaries, research findings, and research methods from papers, which can then be organized into a structured table.

 

 

 

Researchers can propose research questions without providing specific information such as keywords or paper titles, and still match relevant papers from the database. Furthermore, Consensus offers a “Consensus Meter” feature, which visually displays the attitude of the retrieved papers towards the research question.

 

Step 1

 

When a question is posed, it will provide a summary, and the“Consensus meter” indicates whether researchers respond with “Yes”, “Possibly”, or “No” to the question.

 

 

Step 2

 

Consensus automatically screens information sources, prioritizing the display of papers that use randomized controlled trials (RCTs), are published in rigorous journals, or have been highly cited.

 

 

 

 

Explainpaper can read and understand the papers uploaded by users, automatically generating summaries, explaining the text highlighted by users, and providing real-time answers to questions.

 

 

 

The literature recommendation services provided by these tools, which rely on various databases and algorithms, can offer valuable references and support for researchers exploring new fields. Given that the accuracy and comprehensiveness of their searches are influenced by the databases and algorithms used, researchers are encouraged to utilize these tools as a means to broaden their research perspective. Additionally, researchers should be cautious when selecting papers for upload, it is advisable to only upload Open Access resources, to avoid potential copyright infringement.