Improve market-wide Wordstat frontier selection
This commit is contained in:
DCCONSTRUCTIONS
parent
c3a946879b
commit
ffdcaa146f
|
|
@ -3,6 +3,7 @@ import { pool } from "../db/client.js";
|
|||
import { getAnchorReviewContract } from "./anchorReview.js";
|
||||
import { getKeywordCleaningByModelTaskRun, getKeywordCleaningContract } from "./keywordCleaning.js";
|
||||
import { getMarketEnrichmentContract, runMarketEnrichment } from "../market/marketEnrichment.js";
|
||||
import { getMarketFrontierDiscoveryContract } from "../market/marketFrontierDiscovery.js";
|
||||
import { getSeoModelProviderStatusContract, runSeoModelTask } from "../modelProvider/seoModelTaskRuns.js";
|
||||
import { updateProjectOntologyApproval } from "../ontology/projectOntologyRepository.js";
|
||||
|
||||
|
|
@ -325,6 +326,30 @@ async function ensureMarketFrontierDiscoveryReady(
|
|||
return { contract, ready: true };
|
||||
}
|
||||
|
||||
const currentDiscovery = await getMarketFrontierDiscoveryContract(projectId);
|
||||
|
||||
if (
|
||||
currentDiscovery.semanticRunId &&
|
||||
currentDiscovery.provider.mode !== "deterministic_fallback" &&
|
||||
currentDiscovery.probeSeeds.length > 0
|
||||
) {
|
||||
const ready = setWorkflowStatus(
|
||||
setStepStatus(contract, "demand_collection", "running", {
|
||||
detail:
|
||||
`Используем сохранённый market frontier discovery: ${currentDiscovery.probeSeeds.length} probes уже готовы, повторный AI Workspace вызов не нужен.`
|
||||
}),
|
||||
"running",
|
||||
{
|
||||
activeStepId: "demand_collection",
|
||||
reason: null,
|
||||
summary: "Используем сохранённый market frontier discovery и продолжаем Wordstat-добор."
|
||||
}
|
||||
);
|
||||
await saveWorkflow(ready, "running");
|
||||
|
||||
return { contract: ready, ready: true };
|
||||
}
|
||||
|
||||
const modelTaskRun = await runSeoModelTask(projectId, {
|
||||
providerId: "codex_workspace",
|
||||
taskType: "seo.market_frontier_discovery"
|
||||
|
|
|
|||
|
|
@ -300,6 +300,8 @@ const EXPANSION_QUERY_BLOCKLIST = [
|
|||
/(^|\s)(задач|проверок)\.$/i,
|
||||
/(^|\s)(искусственн[а-яёa-z0-9-]*\s+интеллект|ии|ai)\s+(платформ[а-яёa-z0-9-]*|систем[а-яёa-z0-9-]*|автоматизац[а-яёa-z0-9-]*)$/i,
|
||||
/^(платформ[а-яёa-z0-9-]*|систем[а-яёa-z0-9-]*|автоматизац[а-яёa-z0-9-]*)\s+(искусственн[а-яёa-z0-9-]*\s+интеллект|ии|ai)(\s|$)/i,
|
||||
/^(искусственн[а-яёa-z0-9-]*\s+интеллект|ии|ai)(\s+и\s+(llm|нейросет[а-яёa-z0-9-]*|модел[а-яёa-z0-9-]*))?$/i,
|
||||
/^(данные\s+и\s+аналитика|агент\s+ассистент|ai[-\s]?ассистенты\s+и\s+агентная\s+автоматизация)$/i,
|
||||
/^(автоматизац[а-яёa-z0-9-]*|применени[ея]|внедрени[ея]|разработк[аи]|создани[ея])\s+(выбор|оценк[аи]|ценност[ьи])$/i,
|
||||
/^(применени[ея]|внедрени[ея]|разработк[аи]|создани[ея])\s+(данн[а-яё]*|процесс[а-яё]*|задач[а-яё]*)$/i,
|
||||
/[.!?:]/,
|
||||
|
|
@ -311,7 +313,8 @@ const DEMAND_COLLECTION_MAX_BASE_RUNS = 1;
|
|||
const DEMAND_COLLECTION_MAX_MARKET_WIDE_RUNS = 1;
|
||||
const DEMAND_COLLECTION_CONTEXTUAL_PROBE_BUDGET = 8;
|
||||
const DEMAND_COLLECTION_MARKET_WIDE_PROBE_BUDGET = 12;
|
||||
const DEMAND_COLLECTION_MARKET_WIDE_ANCHOR_PROBE_BUDGET = 8;
|
||||
const DEMAND_COLLECTION_MARKET_WIDE_ANCHOR_PROBE_BUDGET = 2;
|
||||
const DEMAND_COLLECTION_MARKET_WIDE_FRONTIER_PROBE_BUDGET = 8;
|
||||
const WORDSTAT_PROBE_MAX_PER_PATTERN_FAMILY = 4;
|
||||
const WORDSTAT_PROBE_MAX_PER_CLUSTER = 8;
|
||||
|
||||
|
|
@ -480,8 +483,8 @@ function getMarketRoleProbeWeight(role: AnchorReviewWordstatQueueItem["marketRol
|
|||
}
|
||||
|
||||
function getProbeSourceWeight(source: WordstatProbeSourceKind) {
|
||||
if (source === "approved_anchor") return 34;
|
||||
if (source === "market_frontier_discovery") return 28;
|
||||
if (source === "market_frontier_discovery") return 42;
|
||||
if (source === "approved_anchor") return 28;
|
||||
if (source === "commercial_demand") return 20;
|
||||
if (source === "market_anchor_expansion") return 14;
|
||||
return 10;
|
||||
|
|
@ -494,11 +497,43 @@ function hasCommercialProbeSignal(phrase: string) {
|
|||
}
|
||||
|
||||
function hasMarketQualifier(phrase: string) {
|
||||
return /(^|\s)(ии|ai|корпоративн[а-яё]*|enterprise|бизнес[а-яё]*|промышленн[а-яё]*|компани[яйи]|предприяти[еяй]|малого\s+бизнеса|отрасл[а-яё]*|закуп|тендер|документооборот|учет|производств|строительств)(\s|$)/i.test(
|
||||
return /(^|\s)(ии|ai|корпоративн[а-яё]*|enterprise|бизнес[а-яё]*|промышленн[а-яё]*|компани[яйи]|предприяти[еяй]|мал[а-яё]*\s+бизнес[а-яё]*|отрасл[а-яё]*|закуп|тендер|документооборот|учет|производств|строительств|беспилот[а-яё]*|бпла|рутинн[а-яё]*\s+задач[а-яё]*)(\s|$)/i.test(
|
||||
phrase
|
||||
);
|
||||
}
|
||||
|
||||
const NON_PRODUCT_PROBE_TARIFF_CONTEXT_PATTERN =
|
||||
/(^|\s)(банк[а-яё]*|банковск[а-яё]*|счет[а-яё]*|счёт[а-яё]*|ип|рко|эквайринг|интернет|связ[ьи]|мобил[а-яё]*|телефон[а-яё]*|оператор[а-яё]*|альфа|втб|сбер|сбербанк|тинькофф|мегафон|билайн|мтс|tele2|360)(\s|$)/i;
|
||||
const PRODUCT_PROBE_TARIFF_QUALIFIER_PATTERN =
|
||||
/(^|\s)(ai|ии|llm|нейросет[а-яё]*|искусственн[а-яё]*\s+интеллект|агент[а-яё]*|ассистент[а-яё]*|автоматизац[а-яё]*|платформ[а-яё]*|сервис[а-яё]*|crm|erp|bpm|workflow|интеграц[а-яё]*|внедрен[а-яё]*|разработк[а-яё]*|корпоративн[а-яё]*|enterprise)(\s|$)/i;
|
||||
const BUSINESS_PLANNING_PROBE_PATTERNS = [
|
||||
/(^|\s)(бизнес[-\s]?план[а-яё]*|создани[ея]\s+бизнес[а-яё]*|открыти[ея]\s+бизнес[а-яё]*)(\s|$)/i,
|
||||
/(^|\s)(бизнес[-\s]?модел[а-яё]*|модел[ьи]\s+для\s+бизнес[а-яё]*|стратеги[яи]\s+для\s+бизнес[-\s]?модел[а-яё]*)(\s|$)/i
|
||||
];
|
||||
|
||||
function hasNonProductProbeIntentMismatch(phrase: string) {
|
||||
const normalizedPhrase = normalizePhrase(phrase);
|
||||
const isTariffPhrase = /(^|\s)(тариф[а-яё]*|расценк[а-яё]*)(\s|$)/i.test(normalizedPhrase);
|
||||
|
||||
if (BUSINESS_PLANNING_PROBE_PATTERNS.some((pattern) => pattern.test(normalizedPhrase))) {
|
||||
return true;
|
||||
}
|
||||
|
||||
if (!isTariffPhrase) {
|
||||
return false;
|
||||
}
|
||||
|
||||
if (NON_PRODUCT_PROBE_TARIFF_CONTEXT_PATTERN.test(normalizedPhrase)) {
|
||||
return true;
|
||||
}
|
||||
|
||||
return (
|
||||
/(^|\s)для\s+(?:мал[а-яё]*\s+|средн[а-яё]*\s+|крупн[а-яё]*\s+)?(бизнес[а-яё]*|предприяти[а-яё]*|компани[а-яё]*)(\s|$)/i.test(
|
||||
normalizedPhrase
|
||||
) && !PRODUCT_PROBE_TARIFF_QUALIFIER_PATTERN.test(normalizedPhrase)
|
||||
);
|
||||
}
|
||||
|
||||
function getWeakGenericProbeReason(phrase: string) {
|
||||
const normalizedPhrase = normalizePhrase(phrase);
|
||||
const words = wordCount(normalizedPhrase);
|
||||
|
|
@ -526,12 +561,12 @@ function getProbeMarketQualityBoost(phrase: string) {
|
|||
(hasMarketQualifier(normalizedPhrase) ? 16 : 0) +
|
||||
(/(^|\s)(купить|заказать|стоимость|цена|тариф|поставщик|подрядчик)(\s|$)/i.test(normalizedPhrase) ? 14 : 0) +
|
||||
(/(^|\s)(для\s+бизнеса|для\s+компани[йи]|для\s+предприяти[яй]|корпоративн[а-яё]*)(\s|$)/i.test(normalizedPhrase) ? 12 : 0) +
|
||||
(/(^|\s)(промышленн[а-яё]*|строительств|закуп|тендер|документооборот|учет|производств)(\s|$)/i.test(normalizedPhrase) ? 10 : 0)
|
||||
(/(^|\s)(промышленн[а-яё]*|строительств|закуп|тендер|документооборот|учет|производств|беспилот[а-яё]*|бпла|мал[а-яё]*\s+бизнес[а-яё]*|рутинн[а-яё]*\s+задач[а-яё]*)(\s|$)/i.test(normalizedPhrase) ? 14 : 0)
|
||||
);
|
||||
}
|
||||
|
||||
function getProbeGenericPenalty(phrase: string) {
|
||||
return getWeakGenericProbeReason(phrase) ? -34 : 0;
|
||||
return getWeakGenericProbeReason(phrase) ? -80 : 0;
|
||||
}
|
||||
|
||||
function getProbePatternFamily(phrase: string) {
|
||||
|
|
@ -592,6 +627,12 @@ function hasAwkwardWordstatProbeSyntax(phrase: string) {
|
|||
/^(заказать|купить|разработк[аи]|создани[ея]|внедрени[ея]|интеграц(?:ия|ии|ию|ией|ие))\s+enterprise\s+(платформ[а-яё]*|систем[а-яё]*|сервис[а-яё]*|решени[ея])/i.test(
|
||||
normalizedPhrase
|
||||
) ||
|
||||
/^(тариф|цена|стоимость)\s+(ops|hub|engine|workspace|platform|cloud)(\s|$)/i.test(normalizedPhrase) ||
|
||||
/(^|\s)(агент|ассистент|роль|workflow|автоматизац[а-яё]*|систем[а-яё]*|платформ[а-яё]*)\s+(ops|hub|engine)(\s|$)/i.test(
|
||||
normalizedPhrase
|
||||
) ||
|
||||
/^(ops|hub|engine)\s+для\s+/i.test(normalizedPhrase) ||
|
||||
/^(операционн[а-яё]*\s+клиент|клиент\s+операционн[а-яё]*|доступа\s+для\s+)/i.test(normalizedPhrase) ||
|
||||
/^(внедрени[ея]|интеграц(?:ия|ии|ию|ией|ие)|разработк[аи]|создани[ея])\s+(сервис[а-яё]*|платформ[а-яё]*|систем[а-яё]*|решени[яе])\s+(ии|ai)\s+(ассистент|агент|платформа|система|сервис)(\s|$)/i.test(
|
||||
normalizedPhrase
|
||||
) ||
|
||||
|
|
@ -631,6 +672,7 @@ function isPlausibleWordstatProbePhrase(phrase: string) {
|
|||
!["n/a", "na", "null", "undefined", "нет данных", "не применимо"].includes(normalizedPhrase) &&
|
||||
!/[{}[\]<>#$%^*=~`/|]/.test(normalizedPhrase) &&
|
||||
!hasAwkwardWordstatProbeSyntax(normalizedPhrase) &&
|
||||
!hasNonProductProbeIntentMismatch(normalizedPhrase) &&
|
||||
!EXPANSION_QUERY_BLOCKLIST.some((pattern) => pattern.test(normalizedPhrase)) &&
|
||||
!hasUnsafeExpansionLatinToken(normalizedPhrase)
|
||||
);
|
||||
|
|
@ -663,6 +705,7 @@ function scoreWordstatProbeCandidate(candidate: WordstatProbeCandidate) {
|
|||
(hasCommercialProbeSignal(normalizedPhrase) ? 12 : 0) +
|
||||
getProbeMarketQualityBoost(normalizedPhrase) +
|
||||
getProbeGenericPenalty(normalizedPhrase) +
|
||||
(hasNonProductProbeIntentMismatch(normalizedPhrase) ? -120 : 0) +
|
||||
(candidate.normalizedFrom?.length ? 4 : 0)
|
||||
);
|
||||
}
|
||||
|
|
@ -852,21 +895,44 @@ function selectMarketWideWordstatProbeSeeds(
|
|||
blockedPhrases
|
||||
);
|
||||
const usedAnchorPhrases = new Set(anchorProbeSeeds.map((seed) => normalizePhrase(seed.phrase)));
|
||||
const blockedAndUsedPhrases = new Set([...blockedPhrases, ...usedAnchorPhrases]);
|
||||
const remainingBudget = DEMAND_COLLECTION_MARKET_WIDE_PROBE_BUDGET - anchorProbeSeeds.length;
|
||||
const anchorFamilyCounts = getWordstatProbeFamilyCounts(anchorProbeSeeds);
|
||||
const frontierBudget = Math.min(
|
||||
DEMAND_COLLECTION_MARKET_WIDE_FRONTIER_PROBE_BUDGET,
|
||||
Math.max(0, DEMAND_COLLECTION_MARKET_WIDE_PROBE_BUDGET - anchorProbeSeeds.length)
|
||||
);
|
||||
const frontierProbeSeeds =
|
||||
frontierBudget > 0
|
||||
? selectWordstatProbeSeeds(
|
||||
candidates.filter((candidate) => candidate.probeSource === "market_frontier_discovery"),
|
||||
wordstat,
|
||||
frontierBudget,
|
||||
new Set([...blockedPhrases, ...usedAnchorPhrases]),
|
||||
{ initialFamilyCounts: anchorFamilyCounts, maxPerFamily: 2 }
|
||||
)
|
||||
: [];
|
||||
const usedFrontierPhrases = new Set(frontierProbeSeeds.map((seed) => normalizePhrase(seed.phrase)));
|
||||
const usedFamilyCounts = getWordstatProbeFamilyCounts([...anchorProbeSeeds, ...frontierProbeSeeds]);
|
||||
const remainingBudget =
|
||||
DEMAND_COLLECTION_MARKET_WIDE_PROBE_BUDGET - anchorProbeSeeds.length - frontierProbeSeeds.length;
|
||||
const derivedProbeSeeds =
|
||||
remainingBudget > 0
|
||||
? selectWordstatProbeSeeds(
|
||||
candidates.filter((candidate) => candidate.probeSource !== "approved_anchor"),
|
||||
candidates.filter(
|
||||
(candidate) =>
|
||||
candidate.probeSource !== "approved_anchor" &&
|
||||
candidate.probeSource !== "market_frontier_discovery"
|
||||
),
|
||||
wordstat,
|
||||
remainingBudget,
|
||||
blockedAndUsedPhrases,
|
||||
{ initialFamilyCounts: anchorFamilyCounts }
|
||||
new Set([...blockedPhrases, ...usedAnchorPhrases, ...usedFrontierPhrases]),
|
||||
{ initialFamilyCounts: usedFamilyCounts }
|
||||
)
|
||||
: [];
|
||||
|
||||
return interleaveWordstatSeedGroups([anchorProbeSeeds, derivedProbeSeeds], DEMAND_COLLECTION_MARKET_WIDE_PROBE_BUDGET);
|
||||
return interleaveWordstatSeedGroups(
|
||||
[frontierProbeSeeds, anchorProbeSeeds, derivedProbeSeeds],
|
||||
DEMAND_COLLECTION_MARKET_WIDE_PROBE_BUDGET
|
||||
);
|
||||
}
|
||||
|
||||
function buildAnchorWordstatProbeSeeds(anchorReview: AnchorReviewContract, suppressedPhrases: Set<string>) {
|
||||
|
|
|
|||
|
|
@ -656,6 +656,144 @@ function getAnchorSeedPhrases(anchorReview: AnchorReviewContract) {
|
|||
}));
|
||||
}
|
||||
|
||||
const SEMANTIC_APPLICATION_FRONTIER_RULES = [
|
||||
{
|
||||
evidencePattern: /bim|digital\s+twin|цифров[а-яё]*\s+двойн|строител|инженер|3d|step|stl|las/i,
|
||||
phrases: [
|
||||
"применение ии в строительстве",
|
||||
"ии в строительстве",
|
||||
"цифровой двойник в строительстве"
|
||||
],
|
||||
title: "Строительство и инженерные объекты"
|
||||
},
|
||||
{
|
||||
evidencePattern: /документооборот|договор|регламент|согласован|процесс|тендер|закуп/i,
|
||||
phrases: [
|
||||
"ии в документообороте",
|
||||
"автоматизация документооборота с ии",
|
||||
"ии автоматизация бизнес процессов",
|
||||
"автоматизация рутинных задач с ии"
|
||||
],
|
||||
title: "Документооборот и бизнес-процессы"
|
||||
},
|
||||
{
|
||||
evidencePattern: /промышлен|производств|iot|телеметр|робот|беспилот|бпла/i,
|
||||
phrases: [
|
||||
"внедрение ии в промышленность",
|
||||
"ии в промышленности",
|
||||
"ии для производства",
|
||||
"ии для беспилотников"
|
||||
],
|
||||
title: "Промышленность и автономные системы"
|
||||
},
|
||||
{
|
||||
evidencePattern: /разработк|api|приложен|код|агент|ассистент|llm/i,
|
||||
phrases: [
|
||||
"разработка ии агентов",
|
||||
"разработка ai ассистентов",
|
||||
"создание ии агентов для бизнеса"
|
||||
],
|
||||
title: "Разработка ИИ-агентов"
|
||||
},
|
||||
{
|
||||
evidencePattern: /enterprise|корпоратив|контур|on[-\s]?prem|private\s+cloud|предприяти/i,
|
||||
phrases: [
|
||||
"внедрение ии в бизнес",
|
||||
"ии для предприятия",
|
||||
"корпоративные ии агенты"
|
||||
],
|
||||
title: "Корпоративное внедрение"
|
||||
},
|
||||
{
|
||||
evidencePattern: /1с|crm|erp|интеграц|api|данн|аналитик|хранилищ/i,
|
||||
phrases: [
|
||||
"ии интеграция с 1с",
|
||||
"ии автоматизация crm",
|
||||
"автоматизация отчетности с ии"
|
||||
],
|
||||
title: "Интеграции и данные"
|
||||
}
|
||||
];
|
||||
|
||||
function getSemanticApplicationSeedPhrases(analysis: SemanticAnalysisRun, anchorReview: AnchorReviewContract) {
|
||||
const semanticText = [
|
||||
...analysis.clusters.flatMap((cluster) => [
|
||||
cluster.title,
|
||||
cluster.targetIntent,
|
||||
...cluster.matchedTerms,
|
||||
...cluster.queryExamples
|
||||
]),
|
||||
...anchorReview.wordstatQueue.flatMap((anchor) => [
|
||||
anchor.phrase,
|
||||
anchor.clusterTitle,
|
||||
anchor.intentType,
|
||||
anchor.marketRole,
|
||||
...anchor.normalizedFrom
|
||||
])
|
||||
].join(" ");
|
||||
const hasAiAutomationContext =
|
||||
/(^|\s)(ии|ai|llm|агент|ассистент|нейросет|искусственн[а-яё]*\s+интеллект|автоматизац)(\s|$)/i.test(semanticText);
|
||||
|
||||
if (!hasAiAutomationContext) {
|
||||
return [];
|
||||
}
|
||||
|
||||
const seeds: Array<{
|
||||
phrase: string;
|
||||
title: string;
|
||||
priority: "high" | "medium" | "low";
|
||||
sourceText: string;
|
||||
evidenceRefs: string[];
|
||||
}> = [
|
||||
{
|
||||
phrase: "ии для бизнеса",
|
||||
title: "Широкий B2B AI спрос",
|
||||
priority: "high",
|
||||
sourceText: "Semantic clusters show AI/agent/automation context; broad B2B probe checks market language before deeper verticals.",
|
||||
evidenceRefs: ["semanticApplication:ai-business"]
|
||||
},
|
||||
{
|
||||
phrase: "ии агенты для бизнеса",
|
||||
title: "AI-агенты для бизнеса",
|
||||
priority: "high",
|
||||
sourceText: "Semantic clusters contain agent/assistant automation; probe validates how buyers phrase agent demand.",
|
||||
evidenceRefs: ["semanticApplication:ai-agents-business"]
|
||||
},
|
||||
{
|
||||
phrase: "автоматизация бизнес процессов с ии",
|
||||
title: "AI-автоматизация процессов",
|
||||
priority: "high",
|
||||
sourceText: "Semantic clusters contain process/workflow automation; probe validates process automation language.",
|
||||
evidenceRefs: ["semanticApplication:ai-process-automation"]
|
||||
},
|
||||
{
|
||||
phrase: "ии для малого бизнеса",
|
||||
title: "Малый бизнес как B2B-сегмент",
|
||||
priority: "medium",
|
||||
sourceText: "Market-wide mode should test SME language when product value is framed as business automation.",
|
||||
evidenceRefs: ["semanticApplication:small-business"]
|
||||
}
|
||||
];
|
||||
|
||||
for (const rule of SEMANTIC_APPLICATION_FRONTIER_RULES) {
|
||||
if (!rule.evidencePattern.test(semanticText)) {
|
||||
continue;
|
||||
}
|
||||
|
||||
for (const phrase of rule.phrases) {
|
||||
seeds.push({
|
||||
phrase,
|
||||
title: rule.title,
|
||||
priority: "high",
|
||||
sourceText: `Semantic cluster evidence matched ${rule.title}; market-wide mode should test this application frontier.`,
|
||||
evidenceRefs: [`semanticApplication:${toId(rule.title)}`]
|
||||
});
|
||||
}
|
||||
}
|
||||
|
||||
return seeds.filter((seed) => isPlausibleProbePhrase(seed.phrase));
|
||||
}
|
||||
|
||||
function getFingerprintProbePhrases(fingerprint: QueryLanguageFingerprint, anchorReview: AnchorReviewContract) {
|
||||
const anchorTerms = unique(
|
||||
anchorReview.wordstatQueue
|
||||
|
|
@ -679,6 +817,7 @@ function getFingerprintProbePhrases(fingerprint: QueryLanguageFingerprint, ancho
|
|||
}
|
||||
|
||||
function buildFallbackFrontiers(input: {
|
||||
analysis: SemanticAnalysisRun;
|
||||
anchorReview: AnchorReviewContract;
|
||||
commercialDemand: SeoCommercialDemandContract | null;
|
||||
fingerprint: QueryLanguageFingerprint;
|
||||
|
|
@ -687,6 +826,7 @@ function buildFallbackFrontiers(input: {
|
|||
const covered = new Set(input.memory.coveredPhrases.map(normalizePhrase));
|
||||
const suppressed = new Set(input.memory.suppressedPhrases.map(normalizePhrase));
|
||||
const seedInputs = [
|
||||
...getSemanticApplicationSeedPhrases(input.analysis, input.anchorReview),
|
||||
...getCommercialSeedPhrases(input.commercialDemand),
|
||||
...getAnchorSeedPhrases(input.anchorReview),
|
||||
...getFingerprintProbePhrases(input.fingerprint, input.anchorReview)
|
||||
|
|
@ -799,6 +939,33 @@ function buildProbeSeedsFromFrontiers(frontiers: MarketFrontierDiscoveryContract
|
|||
return seeds.slice(0, 48);
|
||||
}
|
||||
|
||||
function mergeMarketFrontierProbeSeeds(
|
||||
primarySeeds: MarketFrontierProbeSeed[],
|
||||
fallbackSeeds: MarketFrontierProbeSeed[],
|
||||
memory: MarketResearchMemory
|
||||
) {
|
||||
const blocked = new Set([...memory.coveredPhrases, ...memory.suppressedPhrases].map(normalizePhrase));
|
||||
const seen = new Set<string>();
|
||||
const merged: MarketFrontierProbeSeed[] = [];
|
||||
|
||||
for (const seed of [...primarySeeds, ...fallbackSeeds]) {
|
||||
const normalizedPhrase = normalizePhrase(seed.phrase);
|
||||
|
||||
if (!isPlausibleProbePhrase(normalizedPhrase) || blocked.has(normalizedPhrase) || seen.has(normalizedPhrase)) {
|
||||
continue;
|
||||
}
|
||||
|
||||
seen.add(normalizedPhrase);
|
||||
merged.push({
|
||||
...seed,
|
||||
normalizedPhrase,
|
||||
phrase: normalizedPhrase
|
||||
});
|
||||
}
|
||||
|
||||
return merged.slice(0, 72);
|
||||
}
|
||||
|
||||
function buildModelTaskContract(input: {
|
||||
analysis: SemanticAnalysisRun;
|
||||
anchorReview: AnchorReviewContract;
|
||||
|
|
@ -955,6 +1122,7 @@ function buildFallbackContract(input: {
|
|||
: null;
|
||||
const frontiers = input.analysis
|
||||
? buildFallbackFrontiers({
|
||||
analysis: input.analysis,
|
||||
anchorReview: input.anchorReview,
|
||||
commercialDemand: input.commercialDemand,
|
||||
fingerprint: input.fingerprint,
|
||||
|
|
@ -1058,6 +1226,7 @@ function normalizeModelOutput(
|
|||
const probeSeeds = (Array.isArray(output.probeSeeds) ? output.probeSeeds : [])
|
||||
.map((seed) => normalizeProbeSeed(seed, frontierById, fallback.marketMemory))
|
||||
.filter((seed): seed is MarketFrontierProbeSeed => Boolean(seed));
|
||||
const mergedProbeSeeds = mergeMarketFrontierProbeSeeds(probeSeeds, fallback.probeSeeds, fallback.marketMemory);
|
||||
const contractWithoutActions: Omit<MarketFrontierDiscoveryContract, "nextActions"> = {
|
||||
...output,
|
||||
analystReview: output.analystReview ?? getDefaultAnalystReview(),
|
||||
|
|
@ -1071,10 +1240,10 @@ function normalizeModelOutput(
|
|||
marketMemory: fallback.marketMemory,
|
||||
queryLanguageFingerprint: output.queryLanguageFingerprint ?? fallback.queryLanguageFingerprint,
|
||||
frontiers,
|
||||
probeSeeds,
|
||||
probeSeeds: mergedProbeSeeds,
|
||||
modelTask: fallback.modelTask,
|
||||
rejectedPatterns: Array.isArray(output.rejectedPatterns) ? output.rejectedPatterns : fallback.rejectedPatterns,
|
||||
summary: output.summary || `Market frontier discovery вернул ${frontiers.length} frontiers и ${probeSeeds.length} Wordstat probes.`
|
||||
summary: output.summary || `Market frontier discovery вернул ${frontiers.length} frontiers и ${mergedProbeSeeds.length} Wordstat probes.`
|
||||
};
|
||||
|
||||
return {
|
||||
|
|
@ -1140,6 +1309,7 @@ export async function getMarketFrontierDiscoveryContract(projectId: string): Pro
|
|||
...latest,
|
||||
marketMemory: fallback.marketMemory,
|
||||
modelTask: fallback.modelTask,
|
||||
probeSeeds: mergeMarketFrontierProbeSeeds(latest.probeSeeds, fallback.probeSeeds, fallback.marketMemory),
|
||||
queryLanguageFingerprint: latest.queryLanguageFingerprint ?? fallback.queryLanguageFingerprint
|
||||
} : fallback;
|
||||
}
|
||||
|
|
|
|||
Loading…
Reference in New Issue