MindTrellis: AI-Assisted Interactive Knowledge Graph Construction
Researchers present MindTrellis, an interactive visual system enabling collaborative construction of dynamic knowledge graphs between users and AI. The system addresses challenges faced by knowledge workers in synthesizing information from multiple documents. Current LLM-based systems allow querying but not knowledge organization, while manual tools like mind maps lack intelligent assistance. MindTrellis lets users query the graph for document-grounded information, introduce new concepts, modify relationships, and reorganize hierarchies. The system supports iterative exploration and co-creation of evolving knowledge representations. The paper is available on arXiv under identifier 2604.23129.
Key facts
- MindTrellis is an interactive visual system for collaborative knowledge graph construction.
- Users and AI jointly build a dynamic knowledge graph.
- Users can query the graph for document-grounded information.
- Users can introduce new concepts, modify relationships, and reorganize hierarchies.
- The system addresses limitations of current LLM-based and manual tools.
- The paper is on arXiv with identifier 2604.23129.
- The work targets knowledge workers synthesizing information from multiple documents.
- The process is inherently iterative.
Entities
Institutions
- arXiv