用 Demo 先跑一遍Try the demo first
Demo 不需要准备文件,适合快速了解导入、去重、筛选和导出按钮在哪里。The demo needs no files and helps you find the import, deduplication, screening, and export steps.
第一次接触系统综述的同学可以先用 Demo 熟悉流程;正式项目从数据库导出的 CSV、RIS、NBIB 或 ENW 文件开始。工作台会保留来源、去重、筛选决定和导出记录,方便你和导师、同伴或审稿人一起核对。项目记录默认保存在当前浏览器;需要分享时由你主动导出文件。Students new to systematic reviews can start with the demo. Formal projects usually begin with CSV, RIS, NBIB, or ENW exports from databases. The workspace keeps source, deduplication, screening decisions, and export records connected so you can review them with supervisors, teammates, or peer reviewers. Project records are saved in the current browser by default; you share them by exporting files.
先看懂流程,再换成自己的数据库导出文件。Learn the flow first, then switch to your own database exports.
Demo 不需要准备文件,适合快速了解导入、去重、筛选和导出按钮在哪里。The demo needs no files and helps you find the import, deduplication, screening, and export steps.
从 PubMed、EndNote、Zotero、CNKI、万方、维普或其他数据库导出文件后,在工作台里核对题名、作者、摘要、年份和来源字段。After exporting records from PubMed, EndNote, Zotero, CNKI, Wanfang, VIP, or another database, check title, author, abstract, year, and source fields in the workspace.
完成标题摘要筛选、全文复核和质量评价后,导出 PRISMA 计数、决定表、排除理由和复核材料。After title/abstract screening, full-text review, and quality appraisal, export PRISMA counts, decision tables, exclusion reasons, and review materials.
这些记录帮助你在写方法部分、组会汇报或回应审稿意见时说明过程。These records help you explain the process in methods sections, lab meetings, or reviewer responses.
适合两位同学或两位复核者在不同电脑上各自筛选,再把结果交给项目负责人合并。Use this when two reviewers work on separate computers and send their results back to the project owner.
项目负责人先导出一个起点文件,复核者用它建立相同项目上下文。The owner exports a start file so reviewers can open the same project context.
Reviewer A 或 B 完成筛选后导出自己的决定文件,里面只包含对应复核者的决定。Reviewer A or B exports a decision file after screening; it contains only that reviewer scope.
合并前会检查项目、记录、复核者范围和重复导入状态;通过后刷新冲突和一致性指标。Before merge, the workspace checks project, records, reviewer scope, and duplicate import state, then refreshes conflicts and agreement metrics.
导出文件在本地生成,适合先自己核对,再放入方法附录、团队交接或审稿回复。Exports are generated locally. Review them first, then use them in methods appendices, team handoff, or reviewer responses.
记录项目名称、PRISMA 版本、AI 模式、应用版本和导出时间,方便确认这份导出来自哪个项目。Records project name, PRISMA version, AI mode, app version, and export time so you can identify the project source.
列出导出包中的文件名、媒体类型、字节数和 SHA-256,用来核对一组导出文件是否保持一致。Lists filenames, media types, byte counts, and SHA-256 values so an export set can be checked for consistency.
汇总 PRISMA 计数、筛选决定、双审冲突、质量评价、来源提醒和 AI 使用记录。Summarizes PRISMA counts, screening decisions, dual-review conflicts, quality appraisal, source warnings, and AI-use records.
如果你需要向团队或审稿人说明工具版本,可以使用下面这些标记和命令。If you need to report tool versions to a team or reviewer, use the markers and commands below.
当前公开工作流以人工决定为准,AI 默认关闭。The public workflow treats human decisions as authoritative, and AI is off by default.
AI 建议会记录在独立的 suggestion log 中。只有人工接受或改写后写入 `ScreeningDecision`,才会影响 PRISMA 计数;被拒绝或暂时忽略的建议只作为辅助记录保存。AI suggestions are recorded in a separate suggestion log. They affect PRISMA counts only after a human accepts or edits them into `ScreeningDecision`; rejected or ignored suggestions remain advisory records.
真实 AI 服务需要用户主动配置。默认状态下,工作台不会自动发送文献内容或密钥。A real AI provider requires explicit user setup. By default, the workspace does not send record content or keys automatically.
这些说明帮助你判断什么时候需要人工复核、补充材料或导师确认。These notes help you decide when human review, extra documentation, or supervisor confirmation is needed.
跨设备协作依赖本地 JSON 文件交接。请保存好项目备份,并在分享前检查是否需要脱敏。Cross-device collaboration uses local JSON file handoff. Keep project backups and check whether shared files need de-identification.
PRISMA 计数、质量评价和 GRADE 摘要可以帮助整理材料,最终方法表述、全文判断和证据等级仍由研究团队确认。PRISMA counts, quality appraisal, and GRADE summaries help organize materials. Final methods wording, full-text judgement, and certainty ratings remain the team's responsibility.
当前公开材料来自仓库测试、fixture 和基准文档。真实脱敏项目和用户测试会在完成后再公开说明。Current public materials come from repository tests, fixtures, and benchmark docs. De-identified real-project and user testing results will be described after they are completed.