feat: 完善标签分类与领域场景归一化

src/app/api/discovery/lib/discovery-service.ts

@@ -405,7 +405,8 @@ export async function upsertTags(tags: ProjectInput['tags']) {
       })
       const shouldUpdateNameEn = incomingTag.nameEn && !matchedTag.nameEn
       const shouldUpdateCategory =
-        matchedTag.category === 'FREE_TAG' && inferredCategory !== 'FREE_TAG'
+        matchedTag.category !== inferredCategory &&
+        ['FREE_TAG', 'RESOURCE_TYPE', 'PROTOCOL_INTERFACE'].includes(matchedTag.category)
 
       if (shouldUpdateNameEn || shouldUpdateCategory) {
         const updatedTag = await prisma.tag.update({

src/lib/tag-taxonomy.ts

@@ -23,6 +23,71 @@ export const FIXED_PROJECT_TYPE_TAGS = [
   },
 ] as const
 
+export const DOMAIN_SCENARIO_PRESET_TAGS = [
+  {
+    slug: 'code-dev',
+    name: '开发者工具/代码',
+    nameEn: 'Developer Tools & Coding',
+  },
+  {
+    slug: 'automation-workflow',
+    name: '自动化/工作流/RPA',
+    nameEn: 'Automation, Workflow & RPA',
+  },
+  {
+    slug: 'knowledge-rag',
+    name: '知识管理/检索/RAG',
+    nameEn: 'Knowledge Management, Retrieval & RAG',
+  },
+  {
+    slug: 'education-research',
+    name: '教育/研究资源',
+    nameEn: 'Education & Research Resources',
+  },
+  {
+    slug: 'model-inference',
+    name: '模型训练/推理',
+    nameEn: 'Model Training & Inference',
+  },
+  {
+    slug: 'api-integration',
+    name: '协议/API/集成',
+    nameEn: 'Protocol, API & Integration',
+  },
+  {
+    slug: 'vision-multimodal',
+    name: '计算机视觉/多模态',
+    nameEn: 'Computer Vision & Multimodal',
+  },
+  {
+    slug: 'data-bi',
+    name: '数据分析/BI/可视化',
+    nameEn: 'Data Analytics, BI & Visualization',
+  },
+  {
+    slug: 'security-privacy',
+    name: '安全/隐私',
+    nameEn: 'Security & Privacy',
+  },
+  {
+    slug: 'enterprise-office',
+    name: '企业应用/办公',
+    nameEn: 'Enterprise Applications & Office',
+  },
+  {
+    slug: 'finance',
+    name: '金融',
+    nameEn: 'Finance',
+  },
+  {
+    slug: 'medical-biomed',
+    name: '医疗/生物医药',
+    nameEn: 'Medical & Biomedicine',
+  },
+] as const
+
+export type DomainScenarioPresetSlug = (typeof DOMAIN_SCENARIO_PRESET_TAGS)[number]['slug']
+
 export type FixedProjectTypeSlug = (typeof FIXED_PROJECT_TYPE_TAGS)[number]['slug']
 
 export const FIXED_PROJECT_TYPE_SLUGS = new Set<FixedProjectTypeSlug>(
@@ -117,34 +182,34 @@ const TECH_STACK_SLUGS = new Set([
   'bun',
   'langchain',
   'langgraph',
+  'pytorch',
+  'transformers',
+  'llamaindex',
+  'agent-framework',
+  'vllm',
+  'litellm',
+  'playwright',
+  'chromadb',
+  'gradio',
+  'streamlit',
+  'vuejs',
 ])
 
 const AI_PARADIGM_SLUGS = new Set([
-  'llm',
+  'ai-agents',
+  'multi-agent-system',
   '大语言模型',
   'rag',
-  'transformers',
-  'pytorch',
   'machine-learning',
   'deep-learning',
   'reinforcement-learning',
   'multimodal',
-  'multimodal-ai',
-  'multi-agent-system',
-  'multi-agent',
-  'autonomous-agents',
-  'vllm',
-  'llamaindex',
-  'agent-framework',
-  'ai-agent-framework',
-  'ai-agents',
+  'natural-language-processing',
+  'model-context-protocol',
+  'sdk',
 ])
 
 const PROTOCOL_INTERFACE_SLUGS = new Set([
-  'model-context-protocol',
-  'mcp-protocol',
-  'openai-api',
-  'sdk',
   'api-gateway',
 ])
 
@@ -154,28 +219,39 @@ const PRODUCT_FORM_SLUGS = new Set([
   '桌面应用',
   'vs-code-extension',
   'workflow-automation',
+  'browser-automation',
+  'knowledge-base',
   'automation',
 ])
 
 const DOMAIN_SCENARIO_SLUGS = new Set([
+  ...DOMAIN_SCENARIO_PRESET_TAGS.map((item) => item.slug),
   'data-analytics',
   '数据分析',
   'knowledge-management',
   '知识管理',
   'ai-security',
-  'natural-language-processing',
-  'nlp',
+  'knowledge-graph',
   'computer-vision',
+  'visualization',
+  'enterprise-ai',
+  'chatbot',
 ])
 
-const RESOURCE_TYPE_SLUGS = new Set([
-  'knowledge-base',
-  'markdown',
-  'ai-research',
-  'guide',
-  'tutorial',
-  'paper',
-])
+const RESOURCE_TYPE_SLUGS = new Set<string>([])
+
+const TAG_SLUG_ALIAS_MAP: Record<string, string> = {
+  mcp: 'model-context-protocol',
+  'mcp-protocol': 'model-context-protocol',
+  llm: '大语言模型',
+  'llm-applications': '大语言模型',
+  'autonomous-agents': 'ai-agents',
+  'multi-agent': 'multi-agent-system',
+  'multimodal-ai': 'multimodal',
+  'ai-agent-framework': 'agent-framework',
+  'workflow-orchestration': 'workflow-automation',
+  'web-automation': 'browser-automation',
+}
 
 const TECH_STACK_KEYWORDS = [
   'python',
@@ -210,13 +286,14 @@ const AI_PARADIGM_KEYWORDS = [
   'multi agent',
   '智能体',
   'agent',
+  'mcp',
+  'sdk',
 ]
 
 const PROTOCOL_INTERFACE_KEYWORDS = [
-  'mcp',
+  'api gateway',
+  'gateway',
   'protocol',
-  'api',
-  'sdk',
   '接口',
   '协议',
 ]
@@ -235,6 +312,24 @@ const PRODUCT_FORM_KEYWORDS = [
 ]
 
 const DOMAIN_SCENARIO_KEYWORDS = [
+  'domain',
+  'scenario',
+  'workflow',
+  'automation',
+  'rag',
+  'retrieval',
+  'developer',
+  'coding',
+  'medical',
+  'health',
+  'biomed',
+  'financial',
+  'trading',
+  'enterprise',
+  'office',
+  'api integration',
+  'visualization',
+  'dashboard',
   'security',
   'analysis',
   'knowledge',
@@ -243,30 +338,33 @@ const DOMAIN_SCENARIO_KEYWORDS = [
   'speech',
   'robot',
   'finance',
+  'integration',
   '安全',
   '分析',
   '知识',
+  '检索',
+  '自动化',
+  '工作流',
+  '开发者',
+  '编程',
+  '代码',
+  '协议',
+  '集成',
+  '企业',
+  '办公',
+  '医疗',
+  '医药',
+  '医学',
+  '生物医学',
+  '金融',
+  '交易',
+  '量化',
+  '隐私',
   '语音',
   '视觉',
 ]
 
-const RESOURCE_TYPE_KEYWORDS = [
-  'guide',
-  'tutorial',
-  'paper',
-  'list',
-  'awesome',
-  'docs',
-  'documentation',
-  'resource',
-  '指南',
-  '教程',
-  '论文',
-  '合集',
-  '文档',
-  '资源',
-  '课程',
-]
+const RESOURCE_TYPE_KEYWORDS: string[] = []
 
 const PROJECT_DOC_KEYWORDS = [
   ...RESOURCE_TYPE_KEYWORDS,
@@ -301,22 +399,18 @@ const PROJECT_MODEL_TAG_SLUGS = new Set([
 
 const PROJECT_DOC_TAG_SLUGS = new Set([
   'knowledge-base',
-  'markdown',
-  'ai-research',
 ])
 
 const PROJECT_TOOL_TAG_SLUGS = new Set([
   'ai-development-tool',
   'agent-framework',
-  'ai-agent-framework',
   'ai-agents',
   'mcp',
   'model-context-protocol',
-  'mcp-protocol',
   'cli',
   'workflow-automation',
+  'browser-automation',
   'automation',
-  'openai-api',
   'sdk',
   'web-application',
   '桌面应用',
@@ -358,27 +452,29 @@ export function inferTagCategory(input: {
   nameEn?: string | null
 }): TagCategory {
   const rawSlug = input.slug.trim().toLowerCase()
-  if (isFixedProjectTypeSlug(rawSlug)) {
+  const canonicalSlug = TAG_SLUG_ALIAS_MAP[rawSlug] ?? rawSlug
+
+  if (isFixedProjectTypeSlug(canonicalSlug)) {
     return 'FIXED_PROJECT_TYPE'
   }
-  const normalizedSlug = normalize(rawSlug)
+  const normalizedSlug = normalize(canonicalSlug)
 
-  if (TECH_STACK_SLUGS.has(input.slug)) {
+  if (TECH_STACK_SLUGS.has(canonicalSlug)) {
     return 'TECH_STACK'
   }
-  if (AI_PARADIGM_SLUGS.has(input.slug)) {
+  if (AI_PARADIGM_SLUGS.has(canonicalSlug)) {
     return 'AI_PARADIGM'
   }
-  if (PROTOCOL_INTERFACE_SLUGS.has(input.slug)) {
+  if (PROTOCOL_INTERFACE_SLUGS.has(canonicalSlug)) {
     return 'PROTOCOL_INTERFACE'
   }
-  if (PRODUCT_FORM_SLUGS.has(input.slug)) {
+  if (PRODUCT_FORM_SLUGS.has(canonicalSlug)) {
     return 'PRODUCT_FORM'
   }
-  if (DOMAIN_SCENARIO_SLUGS.has(input.slug)) {
+  if (DOMAIN_SCENARIO_SLUGS.has(canonicalSlug)) {
     return 'DOMAIN_SCENARIO'
   }
-  if (RESOURCE_TYPE_SLUGS.has(input.slug)) {
+  if (RESOURCE_TYPE_SLUGS.has(canonicalSlug)) {
     return 'RESOURCE_TYPE'
   }
 
@@ -386,9 +482,6 @@ export function inferTagCategory(input: {
   const normalizedNameEn = normalize(input.nameEn || '')
   const mergedText = `${normalizedSlug} ${normalizedName} ${normalizedNameEn}`
 
-  if (includesAny(mergedText, RESOURCE_TYPE_KEYWORDS)) {
-    return 'RESOURCE_TYPE'
-  }
   if (includesAny(mergedText, TECH_STACK_KEYWORDS)) {
     return 'TECH_STACK'
   }
@@ -405,7 +498,8 @@ export function inferTagCategory(input: {
     return 'DOMAIN_SCENARIO'
   }
 
-  return 'FREE_TAG'
+  // Free tags are deprecated in this project; default to tech stack for unknown slugs.
+  return 'TECH_STACK'
 }
 
 export function inferProjectTypeSlug(input: {