feat(context): Phase 3 - Context Scoring & Ranking #81

New Issue

cardosofelipe · 2026-01-04T00:51:16Z

cardosofelipe commented

2026-01-04 00:51:16 +00:00

Overview

Implement context scoring algorithms and ranking logic for optimal context selection.

Parent Issue

#61: Context Management Engine

Implementation Tasks

1. Create `scoring/relevance.py`

Create RelevanceScorer class
Implement embedding-based similarity scoring (via Knowledge Base)
Handle contexts without embeddings (fallback to keyword matching)
Cache relevance scores

class RelevanceScorer:
    def __init__(self, mcp_manager: MCPClientManager):
        self.mcp = mcp_manager
        self._cache: dict[str, float] = {}

    async def score(
        self,
        context: BaseContext,
        query: str
    ) -> float:
        """Score context relevance to query using embeddings."""
        ...

2. Create `scoring/recency.py`

Create RecencyScorer class
Implement time-based decay scoring
Make decay rate configurable

class RecencyScorer:
    def __init__(self, decay_hours: float = 24.0):
        self.decay_hours = decay_hours

    def score(self, context: BaseContext) -> float:
        """Score based on recency with exponential decay."""
        age_hours = (datetime.utcnow() - context.timestamp).total_seconds() / 3600
        return math.exp(-age_hours / self.decay_hours)

3. Create `scoring/composite.py`

Create CompositeScorer class
Implement weighted scoring combination
Make weights configurable per context type

4. Create `prioritization/ranker.py`

Create ContextRanker class
Implement parallel scoring using asyncio.gather
Implement greedy selection within budget
Handle priority overrides (priority > 0 forces inclusion)

class ContextRanker:
    async def rank(
        self,
        contexts: list[BaseContext],
        query: str,
        budget: TokenBudget,
        scorer: CompositeScorer
    ) -> list[BaseContext]:
        """Rank contexts by score, respecting budget."""
        ...

Files to Create

backend/app/services/context/
├── scoring/
│   ├── __init__.py
│   ├── relevance.py
│   ├── recency.py
│   └── composite.py
└── prioritization/
    ├── __init__.py
    └── ranker.py

Acceptance Criteria

Relevance scoring uses Knowledge Base embeddings
Recency scoring implements exponential decay
Composite scorer combines weighted scores
Ranker respects budget constraints
Parallel scoring improves performance
Priority overrides work correctly
Unit tests for all scoring algorithms

Dependencies

#69 (Phase 1 - Foundation)
#70 (Phase 2 - Token Budget)

Labels

phase-2, context, backend

## Overview Implement context scoring algorithms and ranking logic for optimal context selection. ## Parent Issue - #61: Context Management Engine --- ## Implementation Tasks ### 1. Create `scoring/relevance.py` - [ ] Create `RelevanceScorer` class - [ ] Implement embedding-based similarity scoring (via Knowledge Base) - [ ] Handle contexts without embeddings (fallback to keyword matching) - [ ] Cache relevance scores ```python class RelevanceScorer: def __init__(self, mcp_manager: MCPClientManager): self.mcp = mcp_manager self._cache: dict[str, float] = {} async def score( self, context: BaseContext, query: str ) -> float: """Score context relevance to query using embeddings.""" ... ``` ### 2. Create `scoring/recency.py` - [ ] Create `RecencyScorer` class - [ ] Implement time-based decay scoring - [ ] Make decay rate configurable ```python class RecencyScorer: def __init__(self, decay_hours: float = 24.0): self.decay_hours = decay_hours def score(self, context: BaseContext) -> float: """Score based on recency with exponential decay.""" age_hours = (datetime.utcnow() - context.timestamp).total_seconds() / 3600 return math.exp(-age_hours / self.decay_hours) ``` ### 3. Create `scoring/composite.py` - [ ] Create `CompositeScorer` class - [ ] Implement weighted scoring combination - [ ] Make weights configurable per context type ### 4. Create `prioritization/ranker.py` - [ ] Create `ContextRanker` class - [ ] Implement parallel scoring using asyncio.gather - [ ] Implement greedy selection within budget - [ ] Handle priority overrides (priority > 0 forces inclusion) ```python class ContextRanker: async def rank( self, contexts: list[BaseContext], query: str, budget: TokenBudget, scorer: CompositeScorer ) -> list[BaseContext]: """Rank contexts by score, respecting budget.""" ... ``` --- ## Files to Create ``` backend/app/services/context/ ├── scoring/ │ ├── __init__.py │ ├── relevance.py │ ├── recency.py │ └── composite.py └── prioritization/ ├── __init__.py └── ranker.py ``` --- ## Acceptance Criteria - [ ] Relevance scoring uses Knowledge Base embeddings - [ ] Recency scoring implements exponential decay - [ ] Composite scorer combines weighted scores - [ ] Ranker respects budget constraints - [ ] Parallel scoring improves performance - [ ] Priority overrides work correctly - [ ] Unit tests for all scoring algorithms --- ## Dependencies - #69 (Phase 1 - Foundation) - #70 (Phase 2 - Token Budget) ## Labels `phase-2`, `context`, `backend`

cardosofelipe referenced this issue

2026-01-04 00:52:21 +00:00

feat(context): Phase 7 - Main Engine & Integration #85

cardosofelipe referenced this issue

2026-01-04 00:54:11 +00:00

feat(mcp): Context Management Engine #61

cardosofelipe closed this issue

2026-01-04 18:53:43 +00:00

Sign in to join this conversation.