Improve ranking for China SaaS localization tasks

lib/query-ranking.test.ts

@@ -0,0 +1,34 @@
+import assert from 'node:assert/strict'
+import test from 'node:test'
+
+import {
+  getLocalizationIntentFitScore,
+  isLocalizationQuery,
+  tokenizeQuery,
+} from './query-ranking.ts'
+
+test('tokenizeQuery removes low-information task words', () => {
+  assert.deepEqual(
+    tokenizeQuery('Localize a SaaS product for launch in China'),
+    ['localize', 'saas', 'product', 'launch', 'china']
+  )
+})
+
+test('isLocalizationQuery recognizes SaaS localization and China launch intent', () => {
+  const query = 'localize a saas product for launch in china'
+  assert.equal(isLocalizationQuery(query, tokenizeQuery(query)), true)
+})
+
+test('localization fit rewards relevant skills and rejects generic coding skills', () => {
+  const relevant = getLocalizationIntentFitScore(
+    'localization',
+    'SaaS localization, internationalization, Chinese locale, and China market entry'
+  )
+  const generic = getLocalizationIntentFitScore(
+    'coding',
+    'General coding patterns for building software products'
+  )
+
+  assert.ok(relevant > 0)
+  assert.ok(generic < 0)
+})

lib/query-ranking.ts

@@ -0,0 +1,66 @@
+const QUERY_STOP_WORDS = new Set([
+  'about',
+  'agent',
+  'agents',
+  'and',
+  'for',
+  'from',
+  'into',
+  'need',
+  'right',
+  'skill',
+  'skills',
+  'that',
+  'the',
+  'this',
+  'use',
+  'using',
+  'want',
+  'what',
+  'when',
+  'with',
+])
+
+export function tokenizeQuery(value: string) {
+  return value
+    .toLowerCase()
+    .split(/[\s+,./:_-]+/)
+    .map((token) => token.trim())
+    .filter((token) => token.length > 2 && !QUERY_STOP_WORDS.has(token))
+}
+
+export function isLocalizationQuery(normalizedQuery: string, queryTokens: string[]) {
+  const tokenSet = new Set(queryTokens)
+  const hasDirectLocalizationTerm =
+    /\b(locali[sz](?:e|ed|ing|ation)|internationali[sz](?:e|ed|ing|ation)|i18n|translat(?:e|ed|ing|ion)|locale|multilingual)\b/.test(
+      normalizedQuery
+    )
+  const hasChinaLaunchContext =
+    /\b(china|chinese|mainland china)\b/.test(normalizedQuery) &&
+    /\b(saas|software|product|launch|market|local)\b/.test(normalizedQuery)
+
+  return (
+    hasDirectLocalizationTerm ||
+    hasChinaLaunchContext ||
+    ['localize', 'localise', 'localization', 'localisation', 'i18n'].some((token) => tokenSet.has(token))
+  )
+}
+
+export function getLocalizationIntentFitScore(category: string, text: string) {
+  const categoryMatch = /\b(localization|localisation|translation|internationalization|internationalisation|i18n|market-entry)\b/.test(
+    category
+  )
+  const positiveMatch = /\b(locali[sz](?:e|ed|ing|ation)|internationali[sz](?:e|ed|ing|ation)|i18n|translat(?:e|ed|ing|ion)|locale|multilingual|china|chinese|market-entry)\b/.test(
+    text
+  )
+  const negativeMatch = /\b(web-crawling|crawler|scraper|browser-automation|presentation|slides?|security|vulnerability)\b/.test(
+    text
+  )
+
+  let score = 0
+  if (categoryMatch) score += 140
+  if (positiveMatch) score += 120
+  if (negativeMatch && !positiveMatch) score -= 170
+  if (!categoryMatch && !positiveMatch) score -= 220
+  return score
+}

lib/registry.ts

@@ -6,21 +6,18 @@ import type { SkillAgentStats, SkillEventStats, SkillOutcomeStats, SkillRecord }
 import { getSkillDecisionProfile } from '@/lib/decision'
 import { getSkillInstallTargets } from '@/lib/install-targets'
 import { getPlatformHints, getSkillQualityProfile } from '@/lib/quality'
+import {
+  getLocalizationIntentFitScore,
+  isLocalizationQuery,
+  tokenizeQuery,
+} from '@/lib/query-ranking'
 import { getSkillAttribution } from '@/lib/skill-attribution'
 import { getSkillSupplyProfile } from '@/lib/supply'
 import { getSkillTrustProfile } from '@/lib/trust'
 import { getUseCasesForSkill, scoreSkillForUseCase, USE_CASES } from '@/lib/use-cases'
 
 const SITE_URL = 'https://www.openagentskill.com'
 
-function tokenize(value: string) {
-  return value
-    .toLowerCase()
-    .split(/[\s+,./:_-]+/)
-    .map((token) => token.trim())
-    .filter((token) => token.length > 2)
-}
-
 const QUERY_TOKEN_ALIASES: Record<string, string[]> = {
   trade: ['trading', 'trader', 'trades', 'finance', 'financial', 'stock', 'stocks', 'market', 'markets', 'portfolio', 'quant', 'backtest'],
   trader: ['trade', 'trading', 'finance', 'financial', 'stock', 'stocks', 'market', 'markets', 'portfolio', 'quant', 'backtest'],
@@ -131,7 +128,7 @@ function skillSearchText(skill: SkillRecord) {
     .toLowerCase()
 }
 
-type QueryIntent = 'finance' | 'presentation' | 'design' | 'coding' | 'sports' | 'research' | 'web' | null
+type QueryIntent = 'finance' | 'localization' | 'presentation' | 'design' | 'coding' | 'sports' | 'research' | 'web' | null
 
 function detectQueryIntent(normalizedQuery: string, queryTokens: string[]): QueryIntent {
   const tokenSet = new Set(queryTokens)
@@ -143,6 +140,10 @@ function detectQueryIntent(normalizedQuery: string, queryTokens: string[]): Quer
     return 'finance'
   }
 
+  if (isLocalizationQuery(normalizedQuery, queryTokens)) {
+    return 'localization'
+  }
+
   if (/\b(presentation|presentations|ppt|pptx|powerpoint|slides?|slide deck|deck|pitch deck|keynote|speaker notes|html slides|visual story)\b/.test(normalizedQuery)) {
     return 'presentation'
   }
@@ -172,8 +173,9 @@ function detectQueryIntent(normalizedQuery: string, queryTokens: string[]): Quer
 
 function getIntentFitScore(intent: QueryIntent, category: string, text: string) {
   if (!intent) return 0
+  if (intent === 'localization') return getLocalizationIntentFitScore(category, text)
 
-  const profile: Record<Exclude<QueryIntent, null>, { category: RegExp; positive: RegExp; negative: RegExp }> = {
+  const profile: Record<Exclude<QueryIntent, null | 'localization'>, { category: RegExp; positive: RegExp; negative: RegExp }> = {
     finance: {
       category: /\b(finance|financial|quant|trading|market|stock|investment|portfolio|fintech|crypto|defi)\b/,
       positive: /\b(finance|financial|quant|quantitative|trade|trades|trader|trading|portfolio|market-data|markets?|stocks?|stock[-_\s]?analysis|equity|crypto|filings?|edgar|sec filing|investor|investment|earnings|10-k|10-q|alpha|factor|backtest|backtesting|risk model|openbb|vectorbt|freqtrade|yfinance|zipline|backtrader|serenity)\b/,
@@ -396,7 +398,7 @@ export function rankSkillsForQuery(
   const rankingQuery = augmentQueryForIntent(query)
   const normalizedQuery = rankingQuery.trim().toLowerCase()
   const compactQuery = normalizedQuery.replace(/[^a-z0-9]+/g, '')
-  const queryTokens = tokenize(rankingQuery)
+  const queryTokens = tokenizeQuery(rankingQuery)
   const expandedQueryTokens = expandQueryTokens(queryTokens)
   const queryIntent = detectQueryIntent(normalizedQuery, queryTokens)
   const isFinanceQueryIntent = /\b(finance|financial|quant|quantitative|trade|trades|trader|trading|invest|investing|investment|portfolio|markets?|stocks?|equity|crypto|filings?|edgar|sec filings?|investor|earnings|10-k|10-q|alpha|factor|backtest|backtesting|risk model)\b/.test(normalizedQuery) ||
@@ -525,7 +527,7 @@ export function getRecommendationReasons(skill: SkillRecord, query: string, scor
   const reasons: string[] = []
   const text = skillSearchText(skill)
   const normalizedQuery = query.trim().toLowerCase()
-  const queryTokens = tokenize(query)
+  const queryTokens = tokenizeQuery(query)
   const matchedTokens = queryTokens.filter((token) => text.includes(token)).slice(0, 4)
 
   if (matchedTokens.length > 0) {

package.json

@@ -10,6 +10,7 @@
     "dev": "next dev",
     "build": "next build",
     "start": "next start",
+    "test": "node --test lib/*.test.ts",
     "seed:popular": "node scripts/seed-popular-skills.ts",
     "indexer:backfill": "node scripts/backfill-high-star-skills.mjs",
     "lint": "eslint ."