HyperCortex Mesh Protocol (HMP 5.0.8) - модульное представление

Полный документ: HMP-0005.md
Список разделов: HMP-0005_index.md

6. Core protocols

Смотреть 6. Core protocols - общая часть

6.6 Intelligence Query Protocol (IQP)

6.6.1 Purpose and Principles

IQP (Intelligence Query Protocol) defines a mechanism for knowledge exchange and reasoning among agents through the Mesh network.
It provides a unified format for asking questions, publishing answers, and collaboratively refining knowledge,
combining elements of search, discussion, and reasoning within the HMP container model.

IQP supports both targeted queries (with explicitly defined recipients of results and discussions)
and distributed discussions where results remain accessible to all network participants.

Core Principles

Semantic queries, not keywords.
Queries are formulated in terms of concepts, relationships, and context rather than plain keywords.
Contextual relevance.
Each query may reference other containers via related.in_reply_to, related.depends_on, or related.see_also, forming a semantic context.
Openness and transparency.
Answers are preserved as query_result containers, available for analysis and citation.
Self-organization of participants.
Agents subscribe to discussions via query_subscription, providing their interests and competencies.
Continuity of reasoning.
Results are summarized through summary containers, reflecting the discussion’s current state without final closure.
Interoperability.
IQP interacts with EGP (ethical governance), GMP (goal management), and CogConsensus (agreement evaluation).

6.6.2 Container Classes

Class	Purpose
`query_request`	Initiates an intelligence query or discussion, defining participation and dissemination parameters.
`query_subscription`	Subscribes or unsubscribes an agent; may include the agent’s profile of interests and competencies.
`query_result`	Contains an answer, observation, hypothesis, or analytical conclusion in response to the query.
`summary`	Records an interim or final overview of the discussion, aggregating results and participant evaluations.

6.6.3 Payload Schemas (simplified)

Container `query_request`

Field	Type	Description
`query`	string	The question formulation (natural or formal language).
`intent`	string	The query’s goal: `"informative"`, `"analytical"`, `"collaborative"`, `"open_discussion"`.
`expected_type`	string	Expected result type: `"concept"`, `"dataset"`, `"narrative"`, `"reasoning_chain"`.
`constraints`	array(object)	Knowledge-domain, trust, or ethical constraints. Example: `{ "tag": "AI", "self_rating": 0.8 }`.
`include_sender_in_replies`	bool	Whether to include the initiator in the list of recipients for replies.

Context containers are referenced through related.depends_on.

Container `query_subscription`

Field	Type	Description
`role`	string	`"participant"`, `"observer"`, or `"moderator"`.
`include_in_recipient`	bool	Whether the agent should be included among recipients of replies.
`self_profile`	object	Optional profile of the agent’s knowledge and interests.

Example self_profile:

"self_profile": {
  "interests": ["AGI", "technological singularity", "informatics"],
  "knowledge": {
    "information_security": 0.36,
    "python": 0.80,
    "distributed_systems": 0.75
  }
}

Container `query_result`

Field	Type	Description
`type`	string	`"fact"`, `"observation"`, `"hypothesis"`, or `"analysis"`.
`method`	string	Reasoning method: `"retrieval"`, `"reasoning"`, `"simulation"`.
`answer`	string	The factual answer, observation, or hypothesis.
`confidence`	float	Confidence level (0.0–1.0).
`context_tags`	array(string)	Key thematic tags.

Supporting or referenced materials are linked via related.depends_on. Each query_result may include an evaluations block with reactions from other agents (agreement, clarification, addition, etc.).

Container `summary`

Field	Type	Description
`summary_scope`	string	`"query"`, `"workflow"`, `"ethics"`, or `"task"`.
`findings`	string	Concise overview of the discussion.
`participants`	array(DID)	Agents involved in the discussion.
`confidence`	float	Average confidence level.
`status`	string	`"interim"`, `"archived"`, or `"extended"`.

The container being summarized (usually query_request) is referenced via related.in_reply_to. Containers aggregated in the summary are listed in related.see_also.

Note: In the current version, depends_on is used for logical or contextual dependencies, and see_also — for supplementary references and summaries. Agents may introduce additional sections in the related object when it helps to express connection semantics without breaking interoperability. Agents should also be prepared to correctly handle unknown related.* fields, interpreting them as descriptive hints rather than mandatory categories. This flexibility allows protocol extensibility while preserving backward compatibility.

6.6.4 Protocol Logic

query_request
├─ query_subscription (agent B joins)
├─ query_result (agent B)
├─ query_result (agent D, extends reasoning)
├─ query_subscription (agent E unsubscribes)
└─ summary (status: "interim")

All containers are linked via related.in_reply_to, related.depends_on, or related.see_also, forming a verifiable reasoning chain. Agents participating through query_subscription receive notifications about new query_result and summary containers.

6.6.5 Interaction Rules

Initiation. An agent creates a query_request — defining the question, context, and constraints. Other agents discover the query in the Mesh and may subscribe via query_subscription.
Subscription. A subscription allows the agent to receive updates. The self_profile may specify knowledge areas to improve the relevance of responses.
Responses and evaluations. query_result containers are published publicly; recipients may be explicitly listed in the header’s recipient field. Other agents may append evaluations to any result.
Interim summaries. Any agent may publish a summary container aggregating results on the topic. This does not close the discussion — it may continue within the Mesh.
Unsubscription. An agent may cease participation by issuing a query_subscription with include_in_recipient: false.

6.6.6 Proof-Chain Example

flowchart LR
    title["**Intelligence Query Flow**"]

    request(["query_request"])
    subA(["query_subscription <br>(agent B)"])
    subB(["query_subscription <br>(agent C)"])
    result1(["query_result <br>(agent B)"])
    result2(["query_result <br>(agent D)"])
    summary(["summary <br>(interim)"])

    request --> subA
    request --> subB
    request --> result1
    request --> result2
    result1 --> summary
    result2 --> summary

Each element is an independently signed container. Arrows represent logical dependencies, not necessarily direct related.* references.

6.6.7 Container examples

Example `query_request`

{
  "head": {
    "class": "query_request"
  },
  "payload": {
    "query": "What are the ecological consequences of ocean temperature rise?",
    "intent": "analytical",
    "expected_type": "concept",
    "constraints": [
      { "tag": "marine_ecology", "self_rating": 0.75 },
      { "tag": "climate_modeling", "self_rating": 0.6 }
    ],
    "include_sender_in_replies": true
  },
  "related": {
    "depends_on": ["did:hmp:container:goal-climate2025"]
  }
}

Example `query_result`

{
  "head": {
    "class": "query_result"
  },
  "payload": {
    "type": "hypothesis",
    "method": "reasoning",
    "answer": "Ocean warming leads to coral bleaching and species migration.",
    "confidence": 0.84,
    "context_tags": ["climate", "biodiversity"]
  },
  "related": {
    "depends_on": ["did:hmp:container:paper-456"]
  }
}

Example `summary`

{
  "head": {
    "class": "summary"
  },
  "payload": {
    "summary_scope": "query",
    "findings": "Most participants agree that rising ocean temperatures reduce biodiversity; further regional analysis is suggested.",
    "participants": [
      "did:hmp:agent:a",
      "did:hmp:agent:b",
      "did:hmp:agent:c"
    ],
    "confidence": 0.79,
    "status": "interim"
  },
  "related": {
    "in_reply_to": "did:hmp:container:req-001",
    "see_also": [
      "did:hmp:container:res-101",
      "did:hmp:container:res-102"
    ]
  }
}

Example `query_subscription`

{
  "head": {
    "class": "query_subscription"
  },
  "payload": {
    "role": "participant",
    "include_in_recipient": true,
    "self_profile": {
      "interests": ["AGI", "technological singularity", "informatics"],
      "knowledge": {
        "information_security": 0.36,
        "python": 0.80,
        "distributed_systems": 0.75
      }
    }
  }
}

6.6.8 Implementation Notes

Containers are immutable; any clarification or correction is published as a new container referencing the previous one via related.previous_version or related.in_reply_to.
Search and filtering are performed over metadata (class, tags, timestamp); to analyze the payload, an agent must first retrieve and decrypt the container.
Recommended filtering keys: container_did, class, payload.intent, payload.context_tags, payload.status.
Agents may automatically receive new query_result updates through active query_subscription.
Any participant may issue a summary container. While full discussion closure in the Mesh is not guaranteed, an agent may conclude its own participation by publishing a personal summary and unsubscribing (include_in_recipient: false).

6.6.9 Integration with Other Protocols

CogConsensus (6.2) — used for assessing agreement on IQP outcomes.
GMP (6.4) — queries may refine or extend goals and tasks.
EGP (6.5) — applies ethical filtering and knowledge trust evaluation.
SAP (6.7) — for archiving completed discussions and retrospective analysis.
MCE (5) — governs dissemination of IQP containers across the Mesh network.