Twinkle HubTwinkle Hub
登入

📌 2026-08-17 新增:⚖️ 台灣法條時光機 — 歷史法條原文 · 修法沿革 · 全文檢索 · 條文關聯圖(935 個 dataset · 35,454+ 筆 row)

查看完整 Changelog →

教甄題目語意+關鍵字檢索

tw_search_teacher_recruit_questions

台灣​專屬MIT

**關鍵字 / Keywords**: 教甄題目 縣市 甄試 選擇題 teacher recruit question 教甄考古題

以自然語言 + (可選) 關鍵字過濾, 檢索**教甄題目**級別.

Corpus: 教師甄選 (各縣市政府), Phase 1 ~5 縣市/聯盟 選擇題逐題抽取
+ Qwen3-Embedding-4B 向量化. 各縣市各科目命題不一, 題型多元.

**與 tw_search_teacher_exam_questions 區隔**:
  - 教檢 (teacher_exam) = 教育部全國共通試題 (取得教師證)
  - 教甄 (teacher_recruit) = 縣市政策性試題 (取得正式教職)

參數:
    query: 自然語言, 例「國小數學分數教學」「特殊教育鑑定」
    stem_contains: 題幹必含此子字串 (硬性 keyword filter)
    county / stage / subject_contains: 同 tw_search_teacher_recruit
    year_from / year_to: 西元年範圍
    limit: top-K (1..100)

回傳 schema:
    {n_corpus, n_returned, query,
     hits: [{paper_id, county, exam_year (民國), exam_year_西元,
             exam_name, stage, subject_name, question_no,
             question_type, stem, options{A,B,C,D}, answer,
             similarity}, ...]}

典型用法:
    tw_search_teacher_recruit_questions("音樂理論基本概念")
    tw_search_teacher_recruit_questions("數學分數", county="taipei", stage="es")
    tw_search_teacher_recruit_questions("特殊教育", subject_contains="特殊")

輸入 schema

{
  "properties": {
    "query": {
      "title": "Query",
      "type": "string"
    },
    "stem_contains": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "title": "Stem Contains"
    },
    "county": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "title": "County"
    },
    "stage": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "title": "Stage"
    },
    "subject_contains": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "title": "Subject Contains"
    },
    "year_from": {
      "anyOf": [
        {
          "type": "integer"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "title": "Year From"
    },
    "year_to": {
      "anyOf": [
        {
          "type": "integer"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "title": "Year To"
    },
    "limit": {
      "default": 20,
      "title": "Limit",
      "type": "integer"
    }
  },
  "required": [
    "query"
  ],
  "title": "tw_search_teacher_recruit_questionsArguments",
  "type": "object"
}

呼叫​方式

配置​好 Twinkle Hub 的 MCP client 後,​agent 會​看到 tw_search_teacher_recruit_questions。直接讓它呼叫​即可,​例如:

# Ask Claude / any MCP client:
請用 tw_search_teacher_recruit_questions 處理 "…"。

# It will call:
tw_search_teacher_recruit_questions(input="…")

尚未​設定 Client?

Claude Desktop 約 3 分鐘即可完成​設定 — 下載 .mcpb 檔案並雙擊​即可,或參閱​文件中的各 Client 安裝​步驟。

查看使用者設定​文件