Flow-guided scientific research and creative evolution generation method, electronic device and computer readable storage medium

By employing a flow-guided method for generating research ideas through evolution, and combining literature knowledge graphs and Monte Carlo tree search, the retrieval path is dynamically adjusted and ideas evolve. This approach addresses the issues of information source diversity and idea quality in the generation of research ideas, and achieves interdisciplinary integration and quality improvement of ideas.

CN122332541APending Publication Date: 2026-07-03DALIAN UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN Β· China
Patent Type
Applications(China)
Current Assignee / Owner
DALIAN UNIV OF TECH
Filing Date
2026-03-31
Publication Date
2026-07-03

AI Technical Summary

Technical Problem

Existing scientific research idea generation technologies suffer from several problems, including a one-way separation between the retrieval and generation processes, limited diversity of information sources, low quality of the initial idea population, and difficulty in effectively balancing the novelty and feasibility of ideas.

Method used

We adopt a flow-guided scientific research idea evolution generation method. By constructing a multidisciplinary literature knowledge graph, combining Monte Carlo tree search and generative reward evaluation, we dynamically adjust the retrieval path and introduce cross-domain information during the idea evolution process to perform cross mutation operations to optimize idea quality.

Benefits of technology

It achieves bidirectional collaborative evolution of the retrieval and generation processes, expands the diversity of information sources, enhances the interdisciplinary integration capability and quality of creative ideas, dynamically improves the novelty and feasibility of creative ideas, and solves the limitations of existing technologies.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122332541A_ABST
    Figure CN122332541A_ABST
Patent Text Reader

Abstract

This invention belongs to the fields of artificial intelligence, natural language processing, information retrieval, and computer-aided scientific research technology. It discloses a stream-guided evolutionary generation method for scientific research ideas, an electronic device, and a computer-readable storage medium. This method constructs a multidisciplinary literature dataset and establishes a literature knowledge graph; explores literature trajectories on the literature knowledge graph, inputs the information of the literature trajectory into a large-scale idea generation model, and outputs the corresponding structured scientific research ideas; performs crossover and mutation operations on the parent population of scientific research ideas; retains the scientific research idea with the highest reward value as the parent population of scientific research ideas for the next crossover and mutation operation, until the final target population of scientific research ideas is obtained. The significant advantage of this invention is that it can generate high-quality scientific research ideas that take into account both novelty and feasibility.
Need to check novelty before this filing date? Find Prior Art