What is Panshi ScienceOne
Panshi ScienceOne is an intelligent research platform jointly developed by the Institute of Automation, Chinese Academy of Sciences, together with multiple institutions and industrialization platforms. Built on a scientific foundation large model, it promotes a new paradigm of multidisciplinary collaborative intelligent research and empowers the entire research process, including hypothesis formulation, solution planning, simulation deduction, experimental validation, and pattern discovery. The core products of Panshi ScienceOne include the S1-Literature assistant and the S1-ToolChain scientific tool scheduling platform. The literature assistant, supported by a rich database of scientific and technical literature, can achieve literature comprehension and automatic review generation, and supports multidisciplinary adaptation; the scientific tool scheduling platform integrates various scientific models and tools through standard protocols, enabling cross-disciplinary data understanding and collaborative task invocation.
Panshi ScienceOne V1.5 adds two major functions: "Panshi Innovation Assessment" and "Panshi Agent Factory." "Panshi Innovation Assessment" can provide assessment support for the innovativeness of research topics and technical solutions, assist with cross-disciplinary topic discovery, research literature submission suggestions, and academic manuscript revision and optimization, forming a research loop from inspiration to impact. "Panshi Agent Factory" can help researchers create and use professional research tools with a lower threshold, integrating core capabilities such as natural language generation, multi-agent collaborative orchestration, sustainable workflows, and agent management, providing efficient and intelligent support for interdisciplinary research.
Main Functions of Panshi ScienceOne
- Literature Assistant (S1-Literature)
- Automatic Literature Review Generation: Based on the largest domestic scientific and technical literature database and various real-time open-source scientific and technical materials, S1-Literature can automatically organize a review skeleton according to brief instructions entered by users, sort through thousands of papers, and generate detailed review content.
- Assisted Close Reading of Papers: Provides tools such as mind maps, citation tracing, research graphs, and key technical path extraction to help users understand and analyze literature more efficiently.
- Multidisciplinary Domain Knowledge Q&A: Based on the scientific foundation large model, the literature assistant supports multidisciplinary domain knowledge Q&A and scientific data interpretation.
- Disciplinary Adaptation and Expansion: Adaptation has been completed for disciplines such as mathematics, physics, and materials, with plans to dynamically expand to full disciplinary coverage.
- Scientific Tool Scheduling Platform (S1-ToolChain)
- Cross-Disciplinary Collaborative Tool Invocation: Enables autonomous collaborative invocation of tools for cross-disciplinary data understanding, scientific computing, and simulation, and supports access to various general and professional scientific models and tools.
- Tool Flow Orchestration and Task Chaining: Through standard protocols for scientific models, agents perform tool flow orchestration and task chaining, and can automatically plan and execute tasks according to research needs.
- Multimodal Scientific Tool Integration: Has integrated nearly 300 tools in mathematics, physics, chemistry, and engineering disciplines, including multimodal scientific data analysis, differential equation solving, discrete optimization, and cross-scale simulation.
- Empowerment Across the Entire Research Process: Panshi ScienceOne can provide support across the entire research process, from "hypothesis formulation—solution planning—simulation deduction—experimental validation—pattern discovery," creating an all-around artificial intelligence research assistant for researchers.
-
Panshi Innovation Assessment: Provide comprehensive assessment for the innovation of research topic selection and technical solutions, assisting interdisciplinary topic discovery, research literature submission recommendations, and academic manuscript revision and optimization.
-
Rocky Intelligent Agent Factory: Integrates core capabilities such as natural language generation, multi-agent collaborative orchestration, sustainable workflows, and intelligent agent management. It can provide efficient and intelligent support for interdisciplinary research, helping researchers quickly develop and deploy professional research tools and accelerate the research process.
How to use Rocky ScienceOne
-
Visit the official website: Visit the Rocky official website https://scienceone.ia.ac.cn/。
-
Register an account: Register an account through the China Science and Technology Cloud Identity Authentication Alliance to obtain free access to the platform.
-
Choose a functional module: Choose functional modules such as the Science Foundation Large Model, Literature Compass, and Toolchain Console according to your needs.
- S1-Literature Literature Assistant
- Generate literature reviews: Users only need to enter a short instruction, such as "Please write a literature review on the topic of...", and the Literature Assistant can automatically organize the review framework, sort through thousands of documents at once, and generate detailed review content.
- Assist in intensive paper reading: The Literature Assistant provides tools such as mind maps, citation tracing, research graphs, and key technical path extraction to help users understand and analyze literature more efficiently.
- S1-ToolChain Science Toolchain Console
- Task planning and execution: Users enter research needs, for example, "Please complete protein sequence completion." The Science Toolchain Console uses the Science Foundation Large Model to identify the research intent, calls the self-developed sequence understanding model to analyze the sequence structure, makes task planning and orchestration, and then calls scientific computing tools and professional models to complete the task.
- Toolflow orchestration: Supports autonomous collaborative invocation of tools such as interdisciplinary data understanding, scientific computing, and simulation, and uses intelligent agents for toolflow orchestration and task chaining through the Science Model Standard Protocol.
Application scenarios of Rocky ScienceOne
- Accelerate the research process: Rocky ScienceOne can empower the entire research process, including hypothesis formulation, solution planning, simulation deduction, experimental verification, and pattern discovery.
- Literature review generation: The S1-Literature Literature Assistant can achieve high-level literature understanding and automatic review generation. Users only need to enter a short instruction, and the Literature Assistant can automatically organize the review framework, sort through thousands of documents at once, and generate detailed review content.
- Assist in intensive paper reading: The Literature Assistant provides tools such as mind maps, citation tracing, research graphs, and key technical path extraction to help users understand and analyze literature more efficiently.