Acceleron: A Tool to Accelerate Research Ideation

Several tools have recently been proposed for assisting researchers duringvarious stages of the research life-cycle. However, these primarily concentrateon tasks such as retrieving and recommending relevant literature, reviewing andcritiquing the draft, and writing of research manuscripts. Our investigationreveals a significant gap in availability of tools specifically designed toassist researchers during the challenging ideation phase of the researchlife-cycle. To aid with research ideation, we propose `Acceleron', a researchaccelerator for different phases of the research life cycle, and which isspecially designed to aid the ideation process. Acceleron guides researchersthrough the formulation of a comprehensive research proposal, encompassing anovel research problem. The proposals motivation is validated for novelty byidentifying gaps in the existing literature and suggesting a plausible list oftechniques to solve the proposed problem. We leverage the reasoning anddomain-specific skills of Large Language Models (LLMs) to create an agent-basedarchitecture incorporating colleague and mentor personas for LLMs. The LLMagents emulate the ideation process undertaken by researchers, engagingresearchers in an interactive fashion to aid in the development of the researchproposal. Notably, our tool addresses challenges inherent in LLMs, such ashallucinations, implements a two-stage aspect-based retrieval to manageprecision-recall trade-offs, and tackles issues of unanswerability. Asevaluation, we illustrate the execution of our motivation validation and methodsynthesis workflows on proposals from the ML and NLP domain, given by 3distinct researchers. Our observations and evaluations provided by theresearchers illustrate the efficacy of the tool in terms of assistingresearchers with appropriate inputs at distinct stages and thus leading toimproved time efficiency.