Charlie Chen
The generation of novel research ideas represents a fundamental challenge in scientific discovery, requiring the identification of unexplored connections, emerging trends, and promising research directions. Knowledge graphs (KGs) have emerged as powerful computational tools for representing and reasoning over scientific knowledge, offering structured approaches to support researchers in ideation processes. This literature review examines the application of knowledge graphs in research idea generation, exploring their construction from scientific literature, methods for identifying research opportunities, and their integration with computational techniques including machine learning and natural language processing. The review specifically focuses on how knowledge graphs enable semantic matching, multi-hop reasoning, link prediction, hierarchical concept dependencies, LLM grounding, personalized exploration, and novelty assessment. By synthesizing current research, this review identifies key methodologies, evaluates their effectiveness, and discusses challenges and future directions in leveraging knowledge graphs to accelerate scientific innovation.
Links:
- Final paper: https://drive.google.com/file/d/1ZwEoUkkGf28vaQ8gq0K1gXGrK9jycaRk/
- Final presentation (slides): https://docs.google.com/presentation/d/142gVAP3frZLOpBrMBZEFmrFOdRUZUDHM/
- Final presentation (video): https://youtu.be/fm_qzjIFM58
