The Web of Science Research Assistant is a responsibly developed, generative AI-powered tool that enhances your research and helps you discover fresh insights faster so you can focus on creating new knowledge. Far more than just a chatbot, our assistant runs alongside you as you work, keeping up with your research needs as they develop. Intelligent discovery helps you effortlessly interpret and explore the literature. Task-based guidance and contextual prompts light up potential paths forward and enable you to complete complex research tasks faster.
1. From the Web of Science homepage, click the Research Assistant tab in the upper-left corner.
2. You will land on the Research Assistant homepage where you can choose how you would like to familiarize yourself with the assistant.
a) Type your own query in the form of a question.
b) Pick a guided task that leads you through each research step. This is perfect for focused questions where you need a clear path to follow.
c) Pick one of our example questions to explore popular topics and see how they're put together. It’s a great way to get inspired and learn how to shape your own research questions.
1. Summary Queries
These are questions that require a written response, such as explaining a concept or providing information on a specific topic (e.g., "What is the role of mitochondrial fusion in cancer progression?").
Response type:
If the query falls under the category of "Overview Queries," we send a curated query to our summarizer tool that leverages the power of generative large language models (LLMs). This tool generates a response using trusted Web of Science content. The generated summary is then provided as the response to your query. This ensures that the information is reliable and up to date. References are included in the response for easy fact-checking.
Example queries:
2. Document Search Queries
These are queries specifically requesting documents, such as research papers or publications on a particular subject or from a particular source (e.g., "What are publications from the University of Pennsylvania in 2015 on the topic of economics?").
Response type:
For "Document Search Queries," we return the 8 most relevant documents as a response to your query. This enables you to access specific research papers or publications that are closely related to your query.
Example queries:
For both types of queries, the response is generated by following these steps:
We start by retrieving articles that exhibit the highest degree of semantic similarity to user query and complement them by adding articles with the utmost relevance through a keyword search. The assistant usually transforms user queries into a structured document search.
Once we have this collection of articles, we organize them in descending order according to their similarity scores to the query. The list of publications goes through a proprietary ranking algorithm. This algorithm further improves the relevancy of the top articles on the list, ensuring that the most relevant references are retrieved.
The assistant uses the abstract content from the top 8 search results to generate a response to the query asked. The response may not leverage all 8 of the references listed. The research assistant chooses which information it requires to provide the best answer to the question.
There are three visualization options in the View more menu next to each document:
1. See related documents (Co-citation map)
2. Analyze this document's references (Enriched cited references map)
3. How this document has been cited (Citing items by classification chart)
Co-citation map
Enter your query in the search bar and Research Assistant will return 8 relevant references. Click the ‘view more’ button on the references and select ‘see related documents. Given a specific paper, Research Assistant will display a visual graph of connections to other papers by citation to identify clusters of closely or co-citing papers or distantly related papers.
Enriched Cited References
Enter your query in the search bar and the Research Assistant will return 8 relevant references. Given a document or documents in a response from Web of Science Research Assistant that have enriched cited references, users can view papers which support, differ, discuss, or include these papers in their background. Users will be able to discover papers that cite these papers in differing contexts to build a more comprehensive view of the research to support their research question or literature review.
How this document has been mentioned
How this document has been mentioned tracks how and where the document has been cited in other academic works. This visualization can give you a sense of its impact and relevance.
Below the results, there are three more visualization options for you to dig deeper into the result set:
1. Documents over time graph
2. Topic map
3. Most cited and connected authors
Documents Over Time Graph
Given an identified topic within a search query, Research Assistant will display a bar chart showing the output of publications (research documents) on that topic by year.
Topic Map
Given an identified topic within a search query, Research Assistant will display a topic map showing relationships between that topic and sub-topics or related topics.
Top Researchers/Researcher Profile Tiles for Top 6 Researchers
When users click on a suggested button or prompt, they should easily access additional relevant research information including top 5 researchers, topic maps, related searches/follow-on questions, citation network map, and publication over time graphs specific to the topic and/or record.
Guided workflows are context-specific prompts to help the user expand or narrow their discovery and complete defined tasks. Choose a guided task on the Research Assistant landing page depending on the research specific task you want to accomplish
To make the most of the assistant and effectively find answers and evaluate resources for your research questions, it's essential to craft well-structured queries. This guide will help you write good queries tailored to the tool's capabilities.
Finding Documents or Document Sets
If your goal is to discover research papers or documents, consider the following guidelines:
Examples:
Papers on biotechnology from the University of California from the years 2020-2023
Open access documents published in Economics in the year 1999
Answering Academic or Scientific Questions
If you need precise answers backed by credible information and references, here's how to create suitable queries:
Examples:
What is the role of mitochondria in cancer?
Difference between global warming and climate change?
By following these guidelines, you can maximize the effectiveness of the Web of Science Research Assistant. Craft precise and well-structured queries to obtain the most accurate and reliable information to support your research and academic effectiveness.