IdeaSynth: Iterative Research Idea Development Through Evolving and Composing Idea Facets with Literature-Grounded Feedback

Abstract: Research ideation involves broad exploring and deep refining ideas. Both require deep engagement with literature. Existing tools focus primarily on idea broad generation, yet offer little support for iterative specification, refinement, and evaluation needed to further develop initial ideas. To bridge this gap, we introduce IdeaSynth, a research idea development system that uses LLMs to provide literature-grounded feedback for articulating research problems, solutions, evaluations, and contributions. IdeaSynth represents these idea facets as nodes on a canvas, and allow researchers to iteratively refine them by creating and exploring variations and composing them. Our lab study (N=20) showed that participants, while using IdeaSynth, explored more alternative ideas and expanded initial ideas with more details compared to a strong LLM-based baseline. Our deployment study (N=7) demonstrated that participants effectively used IdeaSynth for real-world research projects at various ideation stages from developing initial ideas to revising framings of mature manuscripts, highlighting the possibilities to adopt IdeaSynth in researcher's workflows.

• The paper introduces a novel system that employs literature-grounded LLM feedback to enhance iterative research idea development.
• It organizes research concepts into facets on a digital canvas, enabling comprehensive visualization and structured idea exploration.
• Lab and deployment studies demonstrated that IdeaSynth significantly improves idea variation and refinement compared to strong LLM baselines.

"IdeaSynth: Iterative Research Idea Development Through Evolving and Composing Idea Facets with Literature-Grounded Feedback" explores a new tool designed to enhance the research ideation process. Research ideation involves two key activities: broadly exploring ideas and deeply refining them. This process requires significant engagement with academic literature, a gap that existing tools often overlook. These tools focus mainly on the generation of broad ideas without supporting iterative refinement and evaluation.

IdeaSynth Overview:

• Literature-Grounded Feedback: IdeaSynth employs LLMs to provide feedback rooted in academic literature, helping users articulate research problems, solutions, evaluations, and contributions. The integration of literature at this stage assists researchers in grounding their ideas within the existing body of knowledge.
• Idea Facets and Canvas Representation: The system organizes research concepts into facets, represented as nodes on a digital canvas. This visual representation allows users to view different aspects of their ideas and see how they interconnect.
• Iterative Refinement: Researchers can iteratively refine their ideas by creating variations and composing different facets. This process encourages the exploration of alternatives and the expansion of initial concepts with greater detail.

Study Findings:

• Lab Study (N=20): Participants using IdeaSynth explored more alternative ideas and expanded their initial concepts more thoroughly compared to a strong LLM-based baseline. This suggests that IdeaSynth effectively supports both the ideation and refinement stages of research development.
• Deployment Study (N=7): In real-world settings, researchers used IdeaSynth at various ideation stages, from developing initial ideas to revising mature manuscripts. This adaptability highlights its potential utility in integrating into researchers' workflows, offering support across different stages of the research process.

Overall, IdeaSynth appears to address the crucial need for a tool that not only generates research ideas but also aids in their meticulous development and refinement through literature-grounded input. The tool's ability to support iterative and comprehensive exploration makes it a promising addition to research workflows.