{"dataset":{"slug":"scientific-assistant","title":"Scientific AI Research Assistant Platform","description":"The assistant capabilities — grounded (scientific search, object explanation, concept comparison, relationship explanation, evidence chains, provenance- & citation-aware answers, related concepts, reading recommendations, scientific summaries, learning-path generation, cross-domain reasoning, the no-hallucination layer) and architecture-ready (answer modes, RAG interfaces, prompt orchestration, conversation memory, LLM integration). The grounded capabilities run real retrieval over the actual graph and surface only real facts.","version":"1.0.0","lastGenerated":"2026-06-29","license":"CC BY-SA 4.0","entityCount":18,"sources":["nasa"]},"entities":[{"id":"assistant_capability:answer-modes","name":"Answer Modes","type":"assistant_capability","domain":"science","description":"An architecture-ready interface for tuning the depth of an answer to its audience — an educational mode for newcomers, a research mode for depth, an expert mode for concision. The modes are defined; a language-model layer would realise them over the same grounded facts.","entryPath":"/assistant/answer-modes"},{"id":"assistant_capability:citation-aware-answers","name":"Citation-Aware Answers","type":"assistant_capability","domain":"science","description":"Answers linked back to the literature and catalogues behind them, through the platform's persistent identifiers and the ADS literature service. A grounded assistant that shows its references.","entryPath":"/assistant/citation-aware-answers"},{"id":"assistant_capability:concept-comparison","name":"Concept Comparison","type":"assistant_capability","domain":"science","description":"Compares two concepts by the real common ground between them — the entities they both connect to in the graph. Mars and Venus, for instance, share atmospheric escape, climate evolution, and their place in the Solar System. A comparison built from relations, not rhetoric.","entryPath":"/assistant/concept-comparison"},{"id":"assistant_capability:conversation-memory","name":"Conversation Memory","type":"assistant_capability","domain":"science","description":"An architecture-ready interface for remembering the thread of a conversation — the entities discussed, the questions asked — so follow-ups stay in context. The interface is prepared; memory stays on the device, private-first, and holds only references into the graph.","entryPath":"/assistant/conversation-memory"},{"id":"assistant_capability:cross-domain-reasoning","name":"Cross-Domain Reasoning","type":"assistant_capability","domain":"science","description":"Reasons across the graph's domains — following a constellation from its stars to the myths named for it — using the real relations that bridge science, culture, and history. Connections across domains, never invented.","entryPath":"/assistant/cross-domain-reasoning"},{"id":"assistant_capability:evidence-chains","name":"Evidence Chains","type":"assistant_capability","domain":"science","description":"Traces the chain of real relations connecting a claim to its supporting entities — for example, from Edwin Hubble through the expansion of the universe to dark energy. Every link in the chain is a genuine edge in the graph, so the reasoning can be followed and checked.","entryPath":"/assistant/evidence-chains"},{"id":"assistant_capability:learning-path-generation","name":"Learning-Path Generation","type":"assistant_capability","domain":"science","description":"Assembles a learning path through a subject from the platform's curated paths and the real dependencies between concepts — a grounded curriculum drawn from the graph, ordered from foundations to frontier.","entryPath":"/assistant/learning-path-generation"},{"id":"assistant_capability:llm-integration","name":"LLM Integration","type":"assistant_capability","domain":"science","description":"The seam where a language model would join — phrasing the grounded facts the retrieval layer supplies, always constrained to what the graph can back. The interface is defined; the model is future work, and when it comes it will explain, never invent.","entryPath":"/assistant/llm-integration"},{"id":"assistant_capability:object-explanation","name":"Object Explanation","type":"assistant_capability","domain":"science","description":"A grounded explanation of any entity — what it is, how it connects to the rest of the graph, and where the knowledge comes from — assembled entirely from the entity's own description, its relations, and its cited sources.","entryPath":"/assistant/object-explanation"},{"id":"assistant_capability:prompt-orchestration","name":"Prompt Orchestration","type":"assistant_capability","domain":"science","description":"An architecture-ready layer for orchestrating a language model over the grounded retrieval — deciding what to fetch, in what order, and how to assemble it. Defined as an interface; no model is wired in yet.","entryPath":"/assistant/prompt-orchestration"},{"id":"assistant_capability:provenance-aware-answers","name":"Provenance-Aware Answers","type":"assistant_capability","domain":"science","description":"Every grounded answer carries the provenance of the entities it draws on — the sources cited on each, and the review status of the knowledge — so the reader can weigh how far to trust it. No answer without its provenance.","entryPath":"/assistant/provenance-aware-answers"},{"id":"assistant_capability:rag-ready-interfaces","name":"RAG-Ready Interfaces","type":"assistant_capability","domain":"science","description":"The grounded retrieval — search, neighbourhoods, evidence chains — is exposed as a retrieval-augmented-generation interface, so a future language model can ground every response in the real graph. The retrieval is real today; the generation is future work.","entryPath":"/assistant/rag-ready-interfaces"},{"id":"assistant_capability:reading-recommendations","name":"Reading Recommendations","type":"assistant_capability","domain":"science","description":"Recommends what to read next from the platform's own entries, following the real relations outward from where a reader is — a grounded path deeper into a subject.","entryPath":"/assistant/reading-recommendations"},{"id":"assistant_capability:related-concepts","name":"Related Concepts","type":"assistant_capability","domain":"science","description":"Suggests the concepts most closely related to any entity — its real neighbours in the graph. Ask about the main sequence and it surfaces the CNO cycle, the HR diagram, and stellar structure, because those are the entities it actually connects to.","entryPath":"/assistant/related-concepts"},{"id":"assistant_capability:relationship-explanation","name":"Relationship Explanation","type":"assistant_capability","domain":"science","description":"Explains how two entities are related by naming the real relations that link them — 'discovered by', 'orbits', 'is a kind of' — so a connection is never asserted without the graph edge that backs it.","entryPath":"/assistant/relationship-explanation"},{"id":"assistant_capability:scientific-search","name":"Scientific Search","type":"assistant_capability","domain":"science","description":"Semantic, scientific search across every entity in the knowledge graph — a direct route from a question to the real objects, concepts, missions, and people that answer it. Grounded in the graph; it returns only what is really there.","entryPath":"/assistant/scientific-search"},{"id":"assistant_capability:scientific-summaries","name":"Scientific Summaries","type":"assistant_capability","domain":"science","description":"Concise summaries assembled from the entities' own curated descriptions and their key relations — a faithful digest of what the graph holds, with nothing added.","entryPath":"/assistant/scientific-summaries"},{"id":"assistant_capability:no-hallucination-layer","name":"The No-Hallucination Layer","type":"assistant_capability","domain":"science","description":"The design principle beneath the assistant: it surfaces only facts already in the knowledge graph, each with its provenance and a traceable chain of relations. A future language model would phrase these grounded facts, never add to them — so an answer can always be checked against the graph.","entryPath":"/assistant/no-hallucination-layer"}]}