X/LittleWhiteBox
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fix lint warnings and update retrieval modules

Browse Source
This commit is contained in:
bielie
2026-02-09 15:26:43 +08:00
parent 5b9da7aaf4
commit 8131d6a15f
10 changed files with 1761 additions and 763 deletions
Download Patch File Download Diff File Expand all files Collapse all files

181
modules/story-summary/vector/retrieval/metrics.js
View File

@@ -1,5 +1,5 @@
// ═══════════════════════════════════════════════════════════════════════════
// Story Summary - Metrics Collector (v2 - 统一命名)
// Story Summary - Metrics Collector (v3 - Deterministic Query + Hybrid + W-RRF)
//
// 命名规范：
// - 存储层用 L0/L1/L2/L3（StateAtom/Chunk/Event/Fact）
@@ -8,21 +8,44 @@
/**
* 创建空的指标对象
* @returns {object} 指标对象
* @returns {object}
*/
export function createMetrics() {
return {
// Query Build - 查询构建
query: {
buildTime: 0,
refineTime: 0,
},
// Anchor (L0 StateAtoms) - 语义锚点
anchor: {
needRecall: false,
focusEntities: [],
queries: [],
queryExpansionTime: 0,
matched: 0,
floorsHit: 0,
topHits: [],
},
// Lexical (MiniSearch) - 词法检索
lexical: {
terms: [],
atomHits: 0,
chunkHits: 0,
eventHits: 0,
searchTime: 0,
},
// Fusion (W-RRF) - 多路融合
fusion: {
denseCount: 0,
lexCount: 0,
anchorCount: 0,
totalUnique: 0,
afterCap: 0,
time: 0,
},
// Constraint (L3 Facts) - 世界约束
constraint: {
total: 0,
@@ -37,7 +60,7 @@ export function createMetrics() {
inStore: 0,
considered: 0,
selected: 0,
byRecallType: { direct: 0, related: 0, causal: 0 },
byRecallType: { direct: 0, related: 0, causal: 0, lexical: 0 },
similarityDistribution: { min: 0, max: 0, mean: 0, median: 0 },
entityFilter: null,
causalChainDepth: 0,
@@ -50,7 +73,7 @@ export function createMetrics() {
evidence: {
floorsFromAnchors: 0,
chunkTotal: 0,
chunkAfterCoarse: 0,
denseCoarse: 0,
merged: 0,
mergedByType: { anchorVirtual: 0, chunkReal: 0 },
selected: 0,
@@ -93,8 +116,11 @@ export function createMetrics() {
// Timing - 计时
timing: {
queryExpansion: 0,
queryBuild: 0,
queryRefine: 0,
anchorSearch: 0,
lexicalSearch: 0,
fusion: 0,
constraintFilter: 0,
eventRetrieval: 0,
evidenceRetrieval: 0,
@@ -109,6 +135,7 @@ export function createMetrics() {
constraintCoverage: 100,
eventPrecisionProxy: 0,
evidenceDensity: 0,
chunkRealRatio: 0,
potentialIssues: [],
},
};
@@ -116,7 +143,7 @@ export function createMetrics() {
/**
* 计算相似度分布统计
* @param {number[]} similarities - 相似度数组
* @param {number[]} similarities
* @returns {{min: number, max: number, mean: number, median: number}}
*/
export function calcSimilarityStats(similarities) {
@@ -137,8 +164,8 @@ export function calcSimilarityStats(similarities) {
/**
* 格式化指标为可读日志
* @param {object} metrics - 指标对象
* @returns {string} 格式化后的日志
* @param {object} metrics
* @returns {string}
*/
export function formatMetricsLog(metrics) {
const m = metrics;
@@ -150,18 +177,41 @@ export function formatMetricsLog(metrics) {
lines.push('════════════════════════════════════════');
lines.push('');
// Query Build
lines.push('[Query] 查询构建');
lines.push(`├─ build_time: ${m.query.buildTime}ms`);
lines.push(`└─ refine_time: ${m.query.refineTime}ms`);
lines.push('');
// Anchor (L0 StateAtoms)
lines.push('[Anchor] L0 StateAtoms - 语义锚点');
lines.push(`├─ need_recall: ${m.anchor.needRecall}`);
if (m.anchor.needRecall) {
lines.push(`├─ focus_entities: [${(m.anchor.focusEntities || []).join(', ')}]`);
lines.push(`├─ queries: [${(m.anchor.queries || []).slice(0, 3).join(', ')}]`);
lines.push(`├─ query_expansion_time: ${m.anchor.queryExpansionTime}ms`);
lines.push(`├─ matched: ${m.anchor.matched || 0}`);
lines.push(`└─ floors_hit: ${m.anchor.floorsHit || 0}`);
}
lines.push('');
// Lexical (MiniSearch)
lines.push('[Lexical] MiniSearch - 词法检索');
lines.push(`├─ terms: [${(m.lexical.terms || []).slice(0, 8).join(', ')}]`);
lines.push(`├─ atom_hits: ${m.lexical.atomHits}`);
lines.push(`├─ chunk_hits: ${m.lexical.chunkHits}`);
lines.push(`├─ event_hits: ${m.lexical.eventHits}`);
lines.push(`└─ search_time: ${m.lexical.searchTime}ms`);
lines.push('');
// Fusion (W-RRF)
lines.push('[Fusion] W-RRF - 多路融合');
lines.push(`├─ dense_count: ${m.fusion.denseCount}`);
lines.push(`├─ lex_count: ${m.fusion.lexCount}`);
lines.push(`├─ anchor_count: ${m.fusion.anchorCount}`);
lines.push(`├─ total_unique: ${m.fusion.totalUnique}`);
lines.push(`├─ after_cap: ${m.fusion.afterCap}`);
lines.push(`└─ time: ${m.fusion.time}ms`);
lines.push('');
// Constraint (L3 Facts)
lines.push('[Constraint] L3 Facts - 世界约束');
lines.push(`├─ total: ${m.constraint.total}`);
@@ -191,7 +241,8 @@ export function formatMetricsLog(metrics) {
lines.push(`├─ by_recall_type:`);
lines.push(`│ ├─ direct: ${m.event.byRecallType.direct}`);
lines.push(`│ ├─ related: ${m.event.byRecallType.related}`);
lines.push(`│ └─ causal: ${m.event.byRecallType.causal}`);
lines.push(`│ ├─ causal: ${m.event.byRecallType.causal}`);
lines.push(`│ └─ lexical: ${m.event.byRecallType.lexical}`);
const sim = m.event.similarityDistribution;
if (sim && sim.max > 0) {
@@ -210,12 +261,9 @@ export function formatMetricsLog(metrics) {
lines.push('[Evidence] L1 Chunks - 原文证据');
lines.push(`├─ floors_from_anchors: ${m.evidence.floorsFromAnchors}`);
// 粗筛信息
if (m.evidence.chunkTotal > 0) {
lines.push(`├─ coarse_filter:`);
lines.push(`│ ├─ total: ${m.evidence.chunkTotal}`);
lines.push(`│ ├─ after: ${m.evidence.chunkAfterCoarse}`);
lines.push(`│ └─ filtered: ${m.evidence.chunkTotal - m.evidence.chunkAfterCoarse}`);
lines.push(`├─ chunk_total: ${m.evidence.chunkTotal}`);
lines.push(`├─ dense_coarse: ${m.evidence.denseCoarse}`);
}
lines.push(`├─ merged: ${m.evidence.merged}`);
@@ -225,7 +273,6 @@ export function formatMetricsLog(metrics) {
lines.push(`│ └─ chunk_real: ${mt.chunkReal || 0}`);
}
// Rerank 信息
if (m.evidence.rerankApplied) {
lines.push(`├─ rerank_applied: true`);
lines.push(`│ ├─ before: ${m.evidence.beforeRerank}`);
@@ -281,8 +328,11 @@ export function formatMetricsLog(metrics) {
// Timing
lines.push('[Timing] 计时');
lines.push(`├─ query_expansion: ${m.timing.queryExpansion}ms`);
lines.push(`├─ query_build: ${m.query.buildTime}ms`);
lines.push(`├─ query_refine: ${m.query.refineTime}ms`);
lines.push(`├─ anchor_search: ${m.timing.anchorSearch}ms`);
lines.push(`├─ lexical_search: ${m.lexical.searchTime}ms`);
lines.push(`├─ fusion: ${m.fusion.time}ms`);
lines.push(`├─ constraint_filter: ${m.timing.constraintFilter}ms`);
lines.push(`├─ event_retrieval: ${m.timing.eventRetrieval}ms`);
lines.push(`├─ evidence_retrieval: ${m.timing.evidenceRetrieval}ms`);
@@ -299,6 +349,7 @@ export function formatMetricsLog(metrics) {
lines.push(`├─ constraint_coverage: ${m.quality.constraintCoverage}%`);
lines.push(`├─ event_precision_proxy: ${m.quality.eventPrecisionProxy}`);
lines.push(`├─ evidence_density: ${m.quality.evidenceDensity}%`);
lines.push(`├─ chunk_real_ratio: ${m.quality.chunkRealRatio}%`);
if (m.quality.potentialIssues && m.quality.potentialIssues.length > 0) {
lines.push(`└─ potential_issues:`);
@@ -319,14 +370,53 @@ export function formatMetricsLog(metrics) {
/**
* 检测潜在问题
* @param {object} metrics - 指标对象
* @returns {string[]} 问题列表
* @param {object} metrics
* @returns {string[]}
*/
export function detectIssues(metrics) {
const issues = [];
const m = metrics;
// 事件召回比例问题
// ─────────────────────────────────────────────────────────────────
// 查询构建问题
// ─────────────────────────────────────────────────────────────────
if ((m.anchor.focusEntities || []).length === 0) {
issues.push('No focus entities extracted - entity lexicon may be empty or messages too short');
}
// ─────────────────────────────────────────────────────────────────
// 锚点匹配问题
// ─────────────────────────────────────────────────────────────────
if ((m.anchor.matched || 0) === 0 && m.anchor.needRecall) {
issues.push('No anchors matched - may need to generate anchors');
}
// ─────────────────────────────────────────────────────────────────
// 词法检索问题
// ─────────────────────────────────────────────────────────────────
if ((m.lexical.terms || []).length > 0 && m.lexical.atomHits === 0 && m.lexical.chunkHits === 0 && m.lexical.eventHits === 0) {
issues.push('Lexical search returned zero hits - terms may not match any indexed content');
}
// ─────────────────────────────────────────────────────────────────
// 融合问题
// ─────────────────────────────────────────────────────────────────
if (m.fusion.lexCount === 0 && m.fusion.denseCount > 0) {
issues.push('No lexical candidates in fusion - hybrid retrieval not contributing');
}
if (m.fusion.afterCap === 0) {
issues.push('Fusion produced zero candidates - all retrieval paths may have failed');
}
// ─────────────────────────────────────────────────────────────────
// 事件召回问题
// ─────────────────────────────────────────────────────────────────
if (m.event.considered > 0) {
const selectRatio = m.event.selected / m.event.considered;
if (selectRatio < 0.1) {
@@ -341,7 +431,7 @@ export function detectIssues(metrics) {
if (m.event.entityFilter) {
const ef = m.event.entityFilter;
if (ef.filtered === 0 && ef.before > 10) {
issues.push(`No events filtered by entity - focus entities may be too broad or missing`);
issues.push('No events filtered by entity - focus entities may be too broad or missing');
}
if (ef.before > 0 && ef.filtered > ef.before * 0.8) {
issues.push(`Too many events filtered (${ef.filtered}/${ef.before}) - focus may be too narrow`);
@@ -355,19 +445,18 @@ export function detectIssues(metrics) {
// 因果链问题
if (m.event.selected > 0 && m.event.causalCount === 0 && m.event.byRecallType.direct === 0) {
issues.push('No direct or causal events - query expansion may be inaccurate');
issues.push('No direct or causal events - query may not align with stored events');
}
// 锚点匹配问题
if ((m.anchor.matched || 0) === 0) {
issues.push('No anchors matched - may need to generate anchors');
}
// ─────────────────────────────────────────────────────────────────
// 证据问题
// ─────────────────────────────────────────────────────────────────
// 证据粗筛问题
if (m.evidence.chunkTotal > 0 && m.evidence.chunkAfterCoarse > 0) {
const coarseFilterRatio = 1 - (m.evidence.chunkAfterCoarse / m.evidence.chunkTotal);
if (coarseFilterRatio > 0.9) {
issues.push(`Very high evidence coarse filter ratio (${(coarseFilterRatio * 100).toFixed(0)}%) - query may be too specific`);
// Dense 粗筛比例
if (m.evidence.chunkTotal > 0 && m.evidence.denseCoarse > 0) {
const coarseFilterRatio = 1 - (m.evidence.denseCoarse / m.evidence.chunkTotal);
if (coarseFilterRatio > 0.95) {
issues.push(`Very high dense coarse filter ratio (${(coarseFilterRatio * 100).toFixed(0)}%) - query vector may be poorly aligned`);
}
}
@@ -376,7 +465,7 @@ export function detectIssues(metrics) {
if (m.evidence.beforeRerank > 0 && m.evidence.afterRerank > 0) {
const filterRatio = 1 - (m.evidence.afterRerank / m.evidence.beforeRerank);
if (filterRatio > 0.7) {
issues.push(`High rerank filter ratio (${(filterRatio * 100).toFixed(0)}%) - many irrelevant chunks removed`);
issues.push(`High rerank filter ratio (${(filterRatio * 100).toFixed(0)}%) - many irrelevant chunks in fusion output`);
}
}
@@ -395,24 +484,36 @@ export function detectIssues(metrics) {
}
}
// 证据密度问题
// chunk_real 比例（核心质量指标）
if (m.evidence.selected > 0 && m.evidence.selectedByType) {
const chunkReal = m.evidence.selectedByType.chunkReal || 0;
const density = chunkReal / m.evidence.selected;
if (density < 0.3 && m.evidence.selected > 10) {
issues.push(`Low real chunk ratio in selected (${(density * 100).toFixed(0)}%) - may lack concrete evidence`);
const ratio = chunkReal / m.evidence.selected;
if (ratio === 0 && m.evidence.selected > 5) {
issues.push('Zero real chunks in selected evidence - only anchor virtual chunks present');
} else if (ratio < 0.2 && m.evidence.selected > 10) {
issues.push(`Low real chunk ratio (${(ratio * 100).toFixed(0)}%) - may lack concrete dialogue evidence`);
}
}
// ─────────────────────────────────────────────────────────────────
// 预算问题
// ─────────────────────────────────────────────────────────────────
if (m.budget.utilization > 90) {
issues.push(`High budget utilization (${m.budget.utilization}%) - may be truncating content`);
}
// ─────────────────────────────────────────────────────────────────
// 性能问题
if (m.timing.total > 5000) {
// ─────────────────────────────────────────────────────────────────
if (m.timing.total > 8000) {
issues.push(`Slow recall (${m.timing.total}ms) - consider optimization`);
}
if (m.query.buildTime > 100) {
issues.push(`Slow query build (${m.query.buildTime}ms) - entity lexicon may be too large`);
}
return issues;
}