Academic Paper Search — Academic Paper 搜索

搜索API Scholar 搜索

Use this 技能 for paper-first retrieval. Prefer scholar-搜索.mjs when the user needs academic literature, candidate references, DOI clues, citation 签名als, or source verification for scholarly work.

This 技能 is especially good for:

discovering English-language papers from a topic or question finding classic / highly cited papers quickly building an initial literature pool for a review, thesis, proposal, or grant 应用 检查ing whether a paper title, source, year, or DOI is plausible finding official landing pages, publisher pages, repositories, and supporting evidence on the 网页 What this 技能 can do 1) Scholar-based paper retrieval

运行:

node {baseDir}/scripts/scholar-搜索.mjs "查询" node {baseDir}/scripts/scholar-搜索.mjs "查询" -n 10 --year-from 2020 --year-to 2026 node {baseDir}/scripts/scholar-搜索.mjs "查询" --mode review node {baseDir}/scripts/scholar-搜索.mjs "查询" --mode 验证 node {baseDir}/scripts/scholar-搜索.mjs "查询" --json

Returns, when avAIlable:

paper title authors publication summary / venue clues year best verification link cited-by count snippet DOI / DOI URL when 检测ed or enriched likely paper type (review, 系统atic-review, primary-study) 搜索 refinement suggestions verification hints for down流 检查ing

Modes:

short列出 — default; gives a practical reading short列出 review — emphasizes likely review/survey-style papers for literature review 工作流s using heuristic title/snippet 检测ion 验证 — emphasizes title/DOI/source verification for a likely candidate paper or clAIm

Use this first for:

literature discovery related work exploration title verification DOI/source 检查ing identifying representative papers for a topic 获取ting better next-step 搜索 suggestions 2) Evidence-oriented 网页 搜索

运行:

node {baseDir}/scripts/网页-搜索.mjs "查询" node {baseDir}/scripts/网页-搜索.mjs "查询" --json

Use this after Scholar 搜索 to find:

publisher landing pages DOI pages institutional repositories lab/project pages author homepages non-paper evidence related to a re搜索 topic Recommended 工作流 A. Build a candidate paper pool

启动 with Scholar 搜索 using 1-3 focused queries. Example:

node {baseDir}/scripts/scholar-搜索.mjs "large language 模型s higher education" -n 8 --year-from 2021 --mode short列出

Then refine queries with:

synonyms narrower task terms domAIn words method names population or 设置ting constrAInts review / 系统atic review when you need overview papers first B. 验证 promising items

For strong candidates, use Scholar 验证 mode or 网页 搜索 on:

exact paper title "paper title" DOI "paper title" publisher

This often surfaces the official landing page or 仓库 page.

C. Normalize 输出s for down流 use

After retrieval:

deduplicate by title / DOI keep official or publisher links when possible retAIn cited-by counts only as rough influence 签名als, not 质量 guarantees convert the final short列出 into the citation 格式化 the user needs Good 查询 patterns

Use patterns like:

"topic keyword" method "topic keyword" review "topic keyword" 系统atic review "topic keyword" site:doi.org via 网页 搜索 when 验证ing DOI presence exact title queries in quotes for verification

Examples:

node {baseDir}/scripts/scholar-搜索.mjs "intrusion 检测ion deep learning review" -n 10 --year-from 2020 --mode review node {baseDir}/scripts/scholar-搜索.mjs "large language 模型s classroom teaching" -n 10 --year-from 2023 --mode short列出 node {baseDir}/scripts/scholar-搜索.mjs '"Attention Is All You Need"' --mode 验证 node {baseDir}/scripts/网页-搜索.mjs '"Attention Is All You Need" DOI'

环境 variables

The scripts read the first avAIlable key from:

SERPAPI_API_KEY 搜索API_API_KEY

No custom base URL override is exposed in this public edition. The 技能 uses the default SerpAPI 端点 for consistency and 审计ability.

Notes Prefer Scholar 搜索 for literature retrieval; use 网页 搜索 for source verification. Keep 搜索 batches small to avoid rate limits. DOI enrichment may 查询 the public Crossref API when a DOI is not obvious in the 搜索 结果. This 技能 is best used as the front end of a literature 工作流: retrieve → 验证 → deduplicate → 格式化 citations. Review/survey 检测ion is heuristic, so 验证 导入ant clAIms on the destination page. For publication or academic writing tasks, do not treat 搜索 输出 as final truth without 检查ing the destination page.