Theme of inquiry
Can LLMs provide genuinely empathetic therapeutic interaction and support?
A question within its area, explored through 6 lines of inquiry below — each a family of specific questions the research asks.
20 specific questions
- Does full conversation history improve or degrade multi-turn retrieval accuracy?
- How does selective history retrieval improve conversational search accuracy?
- Do dialogue systems need different retrieval strategies for opinions versus factual knowledge?
- Why does selective conversation history outperform including all prior context?
- Why do conversational systems struggle more than static retrieval with ambiguous queries?
- Why does selective context retrieval outperform including all historical information?
- How does multi-turn dialogue improve user satisfaction in search interactions?
35 specific questions
- Does unrestricted reasoning per search step degrade iterative quality over time?
- Why do per-turn thinking budgets matter alongside iterative retrieval depth?
- How should iterative research tasks limit context per reasoning turn?
- Why do per-turn reasoning caps improve iterative search quality?
- What is the optimal balance between search rounds and reasoning depth per round?
- Do single-step retrieval systems with sophisticated synthesis qualify as deep research?
- Why do deep research agents outperform retrieval augmented generation systems?
50 specific questions
- Can knowledge graph structure be exploited for efficient multi-hop retrieval?
- Can graph-based retrieval with knowledge graphs scale to multi-hop reasoning?
- How do community-based summaries differ from retrieval-based traversal in knowledge graph RAG?
- How do taxonomy-based retrieval scaffolds improve model performance at inference time?
- Can inference-time query decomposition replace pre-built knowledge graph structures?
- When should you use knowledge graphs instead of semantic vector retrieval systems?
- Can query-time logic graphs match the efficiency of pre-built knowledge graph indexing?
58 specific questions
- How do vector embeddings fail to capture task-relevant document relationships?
- Why do embedding-based retrieval systems fail on vocabulary mismatch?
- Can vector embeddings measure task relevance instead of semantic similarity?
- Why do semantic similarity and task relevance diverge in vector search results?
- What makes vector embeddings fail on single-hop semantic relevance queries?
- Can embedding-based retrieval alone solve the causal relevance problem?
- What mathematical limits constrain embedding-based retrieval systems?
56 specific questions
- When does long-context LLM reasoning fail where structured retrieval succeeds?
- Can adaptive per-step decisions outperform uniform retrieval policies across different reasoning tasks?
- How should retrieval systems handle multi-hop reasoning and iterative information needs?
- Can parallel retrieval chains avoid the context consumption problem?
- How does query decomposition reduce retrieval costs at inference?
- What limits exist on retrieval budget during inference?
- What makes proactive tool retrieval better than single-round semantic matching?
49 specific questions
- Can adaptive retrieval triggered by model uncertainty improve RAG reliability?
- Should retrieval be triggered always or only for difficult questions?
- What causes the retrieval-augmented generation to fail in practice?
- How does response content compare to model confidence as a retrieval trigger?
- How should retrieval systems decide when to fetch new information?
- Could eliminating retrieval entirely work better than shifting the burden?
- Can long-context models replace retrieval-augmented generation systems?