Holmes

Holmes is SocioProphet's open language intelligence fabric.

It is built to outgrow assistant-grade discovery: classical NLP, neural NLP, semantic search, retrieval, knowledge graphs, foundation-language services, guardrails, evals, and investigative agentic discovery under one governed product surface.

Product thesis

Component NLP annotates. Holmes investigates.

Holmes is not a chatbot wrapper, a loose model zoo, or a domain NLP repo. It is the governed language layer above search, evidence, retrieval, casefiles, semantic graphs, tools, models, evals, and agents.

Product family

• Holmes: language intelligence fabric.
• Sherlock Search: discovery, retrieval, evidence search, and investigation engine.
• 221B: casefile and workspace surface.
• Mycroft: model routing, policy intelligence, and strategic model/service selection.
• Moriarty Bench: adversarial eval and red-team harness.
• Irene Shield: privacy, masking, identity-sensitive redaction, and sensitive-context handling.
• The Canon: curated evidence corpus, provenance records, accepted facts, and source trust.
• Deduction Engine: synthesis, contradiction detection, claim extraction, fallacy analysis, and reasoning workflows.

Layer stack

1. Ingestion: documents, web, OCR handoff, transcripts, tables, metadata, language detection.
2. Linguistic primitives: tokenization, lemmatization, POS, morphology, parsing, NER, normalization.
3. Rule and table techniques: matchers, gazetteers, taxonomies, table extraction, rule-based relation extraction.
4. Classical ML NLP: classifiers, clustering, topic models, sentiment baselines, calibration, explainability.
5. Neural NLP: transformers, embeddings, rerankers, span extraction, relation extraction, multilingual encoders.
6. Foundation language services: extraction, summarization, generation, translation, RAG answering, long-context analysis, tool planning.
7. Retrieval and knowledge: sparse/dense/hybrid retrieval, vector stores, GraphBrain, semantic-serdes, ontogenesis, Slash Topics, Sherlock Search.
8. Topic-model training support: topic seeds, topic boundaries, candidate labels, taxonomy candidates, topic-pack generation receipts, and Slash Topics training references.
9. Guardrails and governance: PII checks, source provenance, prompt-injection checks, policy gates, eval gates, factsheets, promotion records.
10. Agent and tool orchestration: tool contracts, agent identity, sessions, memory, MCP/A2A, execution traces, model routing.

Slash Topics training role

Holmes supports Slash Topics by producing governed topic-model training artifacts, not opaque topic labels.

For Slash Topics, Holmes emits or prepares topic seeds, positive/negative/adjacent/ambiguous boundary evidence, candidate labels, topic taxonomy candidates, eval slices, and replayable topic-pack generation receipts. Slash Topics owns /topic pack semantics and policy membranes; Holmes owns language evidence, candidate generation, model-training support, and promotion evidence required before a topic model or topic pack can become stable.

Proof-claim reasoning role

Holmes reasoning-layer ownership is bounded to claim Propose -> Explain -> Verify contracts. Holmes emits Claim candidates plus ExplanationTrace, optional ProofCertificate, and ContradictionReport artifacts for downstream policy evaluation. Holmes does not make policy admission decisions.

Reference: docs/PROOF_CLAIM_CONTRACT.md

NLP component alignment

Holmes explicitly covers these component families:

• basic primitives;
• advanced primitives;
• rule techniques;
• classical ML;
• neural NLP;
• transformers;
• foundation-language services;
• retrieval and knowledge;
• guardrails and governance;
• agent and tool orchestration.

The alignment contract is documented in docs/NLP_COMPONENT_ALIGNMENT.md. That document is the lower-layer NLP map for Holmes, nlplab, Sherlock Search, Slash Topics, and the platform runtime.

Repo role

This repo is the Holmes product surface and integration spine.

Normative cross-surface standards live in SocioProphet/functional-model-surfaces. Runtime service deployment should graduate into SocioProphet/prophet-platform when contracts and smoke tests are stable. Linux-native NLP lab execution belongs in SociOS-Linux/nlplab. SourceOS carries clients and signed service references through SourceOS-Linux/sourceos-model-carry.

Contract responsibility catalog

• Holmes: proof-claim reasoning contracts and verification/explanation artifacts.
• Sherlock: retrieval and evidence discovery used as inputs to reasoning.
• Slash Topics: governed /topic pack semantics, topic membranes, and topic-model surfaces trained from Holmes evidence artifacts.
• GAIA: graph/world grounding inputs consumed by Holmes reasoning adapters.
• Agentplane: governed execution after policy outcomes are decided.
• Policy Fabric: admission decisioning after Holmes evidence, proof, contradiction, and training artifacts are produced.

Initial validation

make validate

Expected result:

OK: Holmes contracts validated