(Tentative)
| Function | Reference Model | Input/Output Data |
| SubAIMs | JSON Metadata | Profiles |
Function
The User State Refinement (PGM‑USR) AIM
- Receives evolving indicators of the User’s activity, engagement, personal status, and contextual cues coming from Context Capture (baseline User State), ASR/VSR, and Domain Access (DAC).
- Transforms these heterogeneous signals into:
(a) a unified User State snapshot suitable for downstream orchestration, and
(b) expressive guidance that informs Personality Alignment and A‑User Formation. - Keeps A‑User behaviour sensitive to the User’s evolving state, supporting emotionally aligned interaction, expressive coherence, and context‑aware orchestration across AIMs.
Reference Model
Figure 1 gives the Reference Model of the User State Refinement (PGM-USR) AIM.

Figure 1 – Reference Model of User State Refinement (PGM-USR) AIM
USR consumes (1) baseline User State and scene descriptors from Context Capture, (2) DAC outputs (notably the User Context Guide UCG), (3) ASR/VSR updates to audio and visual anchors, and (4) Interaction History / Enhanced Response. USR refines Engagement, Personal Status, Motivation, and Disposition;
USR generates User State Status (USS), Expressive Guidance (ESG), and a UR‑Prompt for the next LLM interaction—aligning personality and formation with the refined state.
Input/Output Data
Table 18 gives Input and Output Data of User State Refinement (PGM-USR) AIM.
Table 18 – Input and Output Data of User State Refinement (PGM-USR) AIM
| Input | Description |
| User Context Guide | Refined intent, comfort indicators, contextual priorities, interaction style, trace. |
| Enhanced Response | First output from Basic Knowledge |
| User State Directive | Instructions from A‑User Control specifying refinement goals, urgency level, or override conditions. |
| Output | Description |
| Expressive State Guide | Modulation hints for Personality Alignment (tone/style/verbosity + safety & override flags), with provenance. |
| UR-Prompt | Second Prompt to Basic Knowledge |
| User State Status | Current engagement, affective tone, and override flags for A-User Control consumption.. |
SubAIMs
The following SubAIMs may be used to implement User State Refinement
| SubAIM | Function | Inputs | Outputs | To |
| ESM – Entity State Merger | Merge baseline Entity/User State with refinements from UCG and prior USS; ensure consistency before recalibration. | Baseline Entity State; UCG; Interaction History; User State Directive | Merged Entity State | ENG |
| ENG – Engagement Recalibrator | Compute engagement level and mode using UCG comfort indicators, cadence history, and directives. | Merged Entity State; UCG; Interaction History; User State Directive | Engagement Metrics | PST |
| PST – Personal Status Refiner | Update affective tone, motivation, and disposition by blending UCG semantics with engagement metrics and Enhanced Response deltas. | Engagement Metrics; UCG; Enhanced Response; User State Directive | Refined Personal Status | USS |
| USS – User State Synthesizer | Consolidate Entity State, engagement, and personal status into authoritative User State Status with provenance. | Refined Personal Status; Engagement Metrics; UCG; Enhanced Response | User State Status (USS) | ESG / URP |
| ESG – Expressive State Guide Composer | Generate tone, style, verbosity, and safety overrides for PAL based on USS and UCG. | USS; UCG; Enhanced Response | Expressive State Guide (ESG) | PAL |
| URP – UR-Prompt Assembler | Build structured UR-Prompt aligned with UR-Prompt Plan, embedding refined state and expressive cues for next LLM cycle. | USS; ESG; UCG; Enhanced Response | UR-Prompt | BKN |
Table 2 describes the informative SubAIMs that have been identified for PGM-PRC.
Table 2 – Informative SubAIMs
| Term | Definition |
| A‑User Control | Supervisor that issues User State Directives (USD), manages Overseer/Micromanager modes, and records provenance/overrides. |
| Baseline Entity State | Initial snapshot emitted by Context Capture before USR refinement; includes identifiers, focus, intent hints, and scene anchors. |
| Disposition | Refined stance/attitude of the User derived by USR (e.g., collaborative, directive, observational). |
| Engagement Level | User activity intensity: {low |
| Engagement Mode | Interaction type: {passive |
| Enhanced Response | Outcome‑driven feedback artefact synthesised after a cycle; provides guidance deltas and trace for the next USR update. |
| Expressive State Guide | Modulation parameters (tone, style, verbosity, safety overrides) produced by USR and consumed only by PAL. |
| Intent Rationale | A section of UR-Prompt that summarizes the refined user intent and disposition and may include natural language guidance for the next response (e.g., tone, verbosity, focus). |
| Interaction History | Session memory of prior USR operation, cadence features, directives, and outcomes that USR uses for recalibration. |
| Micromanager Mode | A‑User Control mode that can override USR sequencing/parameters on urgency/failure/ambiguity. |
| Overseer Mode | Default A‑User Control mode issuing symbolic USD and monitoring outputs without intrusive overrides. |
| Provenance / Trace | Origin, timestamps, linked IDs (UCG, USD, prompts) that USR preserves in USS, ESG, and UR‑Prompt. |
| Refined Personal Status | USR‑enriched affective tone, motivation, and disposition produced after blending UCG semantics and signals. |
| Safety Overrides | Constraints derived from comfort indicators (e.g., avoid depth cues, reduce complexity) that PAL uses to adapt expressive framing. |
| UR‑Prompt | Prompt assembled by USR for the next LLM cycle; references UR‑Prompt Plan (UPP) and embeds state/guidance. |
| UR‑Prompt Plan | Schema defining the structure and required fields of UR‑Prompt across cycles. |
| User Context Guide | DAC‑produced guide with refined intent, comfort indicators, priorities, interaction style, and trace; a key USR input. |
| User State Directive | Control signal from A‑User Control specifying refinement goals, urgency, or overrides for USR. |
| User State Status | USR output describing engagement level, affective tone, disposition, override report, and trace for use by AUC. |
JSON Metadata
https://schemas.mpai.community/PGM1/V1.0/AIMs/UserStateRefinement.json
Profiles
No Profiles.