04/09/2024 | Press release | Distributed by Public on 04/09/2024 14:14
The Web of Science™ Research Assistant, a generative AI-powered tool built with and for researchers, is now available to the global research community.
After extensive collaboration and testing with the research community that began in December 2023, we are pleased to announce the global launch of the Web of Science Research Assistant. The research assistant combines the power of generative artificial intelligence (AI), the Web of Science knowledge graph and an all-new chat interface to provide researchers with a groundbreaking way of querying and interacting with over 120 years of trusted data from the Web of Science Core Collection™. Researchers at any stage in their career can rely on the research assistant to not only streamline search, but also help them achieve more involved research tasks and goals faster.
A strong set of AI principles and close collaboration with the research community are central to the development of innovative research tools at Clarivate™. To ensure that the Web of Science Research Assistant meets the demands of modern research and addresses valid concerns about the use of AI in academia, Clarivate partnered with librarians and researchers around the world early in the development process. Development partners worked with the Web of Science product team for over nine months to shape, test and refine features.
As one of our beta testers, Li-Ling Ou, Assistant Administrator, Division of Knowledge Service at the National Cheng Kung University Library, explains: "We chose to engage in the Web of Science development partner program due to the database's reputation for credibility and reliability. The high quality articles available through Web of Science, coupled with the opportunity to experience a trial period, provided us with a positive and valuable experience."
Development partners help ensure that the generative AI-powered assistant meets rigorous standards for quality, accuracy and privacy. To minimize hallucinations and bias, the assistant leverages only the trusted, high-quality data in the Web of Science Core Collection. The tool delivers carefully curated data from editorially selected sources, providing researchers with confidence that they are building on a reliable research foundation. By hosting the large language model (LLM) on a private cloud, Clarivate guarantees that data is not shared with other parties or used for training internal or external models.
Researchers consult the literature with a wide range of aims and goals in mind. Today they encounter an immense corpus of previously published research. Surfacing relevant results in response to a query remains a basic database feature, but researchers need smarter tools that help them make sense of results and interpret the literature more efficiently.
Figure 1: Web of Science Research Assistant offers intelligent discovery and task-based guidance
The new capabilities of the Web of Science Research Assistant are designed to tackle those challenges. The assistant can respond to a range of different queries, no matter the language. It offers a brand-new way of interacting with Web of Science Core Collection that enables researchers to:
Figure 2: Topic maps reveal related subject areas
Figure 3: Documents over time graphs reveal publication trends
The Web of Science Research Assistant leverages responsible AI and a bespoke knowledge graph to extract information and visualizations from a trusted corpus of literature and delivers those valuable insights to researchers. It is powered by the Clarivate Academic AI Platform, which serves as a technology backbone, enabling accelerated and consistent deployment of AI capabilities across the Academia & Government portfolio of solutions at Clarivate.
To maximize the strength of generative AI for research use cases, our data scientists optimized the content preparation and search indexing towards content in the scientific and academic research domain, namely the rich Web of Science dataset. Each query is first passed through a proprietary router and parsed into searchable terms. Then results are surfaced via retrieval augmented generation (RAG) to ensure the knowledge extracted is limited to only the trusted content in the Web of Science Core Collection.
If the assistant cannot respond to a question, it will automatically attempt to rephrase it into a research question it can answer based on reliable content. In the generated response, every statement is referenced to its source document to ensure you can verify the answer side-by-side with the underlying research.
Figure 4: View referenced document details alongside generated responses
When layered on the interconnected network of information in the Web of Science knowledge graph, AI technology helps users perform more sophisticated queries and deepen their understanding of the literature. By taking effort away from the researcher and putting it into the search itself, the research assistant makes it easy to discover connections between content, people, and institutions.
The Web of Science Research Assistant is not only a partner to help answer questions, it also helps researchers complete complex research tasks faster. Guided walk-throughs are available to help researchers carry out several common tasks, including understanding a new subject area, performing a literature review and looking for an appropriate, vetted journal to submit a manuscript to.
Through a combination of step-by-step guides and contextual prompts delivered throughout a session, the research assistant leads users to key papers, people, trends, and related topics in a field. It also lights up potential paths forward so researchers can head down new, unconsidered avenues of exploration faster.
Figure 5: Contextual prompts provide options for exploring an area further
Clarivate is focused on implementing next-generation capabilities that remove roadblocks in the research and publishing process and ultimately contribute to the betterment of research. Far more than a chatbot, the new Web of Science Research Assistant runs alongside researchers as they work, improving their process and keeping up with their research needs as they develop. By cutting through the complexity of data, the assistant helps researchers to research smarter so they can think forward.
Learn more in our upcoming webinar.
Interested in exploring a trial for your institution?