Academic Paper Summaries: a Tokeny case study
Batch-read academic PDFs, extract core insights and methodologies, and generate literature reviews. Follow the recorded task, execution screenshots, actual files and verification notes.
By the Tokeny team · Windows · Tokeny 1.7.8 · gpt-6-sol
Prepare the inputs
Create a separate workspace in Tokeny, download and unpack the sample inputs, then select a model that supports tools. The sample uses synthetic data or the public sources named in the article.
- paper-a.pdf
- paper-b.pdf
Send the task to Tokeny
The original task was executed in Chinese. Copying these instructions reproduces the same task; the results below include an English report.
读取两份本地演示研究论文 PDF，提取研究问题、方法、结果和局限，生成文献对照表与综述。它们是合成教学材料，不能称为真实学术证据。 交付文件：literature.csv、review.md。 验收要点： - 两份 PDF 都被读取 - 保留样本量与局限 - 明确是合成论文 这是使用合成演示输入或指定公开资料的真实操作案例。请先读取工作空间输入，再实际生成文件并自检。所有写入、脚本与临时服务限于当前工作空间；临时服务只监听 127.0.0.1，执行后关闭；不触达外部收件人、不发布、不付款、不更改系统设置，不调用子智能体。需要公开资料时可读取给定 URL。只启用完成任务必需的工具。另写 report.md（中文）和 report.en.md（英文），简洁、完整，包含输入、处理步骤、真实结果、检查方式和局限；失败或未验证部分如实记录，不能宣称成功。最终回复用中文列出产物、关键结果与待核实项。 排版检查：Word 主标题仅出现一次，标题为黑色且与后续正文或表格保持同页；Excel 表头和长文字不能被窄列截断，合理设置列宽、换行、行高；HTML 在 390px 视口不能整页横向溢出。报告没有固定字数要求，不要为字数反复改写。
The recorded run used an isolated demonstration workspace. Review the commands and file paths before approving operations in your own workspace.
What actually ran
The task was submitted through the Tokeny input box. An online model read the inputs and executed tools. The inputs are two synthetic teaching papers used to exercise PDF reading and synthesis.
Results and verification
Inputs
The two local PDFs in the workspace root were read: paper-a.pdf and paper-b.pdf. Each file is one page. Both explicitly identify themselves as synthetic teaching papers and state that the data are constructed/synthetic rather than real customer data.
Processing Steps
- Parsed both PDFs locally with the installed
pypdfpackage. - Wrote extracted text to
.tmp/tokeny/pdf_extract/for processing verification. - Extracted the research question, method, sample size, result, and limitations from each paper.
- Generated
literature.csvandreview.md. - Checked the CSV fields, key numeric values, limitations, and the synthetic-evidence disclaimer.
Actual Results
- Paper A: 20 synthetic students, 10 per group, over 2 weeks; means of 78/100 for spaced practice and 70/100 for massed practice; limitations: constructed data, small sample, and no causal conclusion.
- Paper B: 12 synthetic learners, 6 per group, over 1 week; means of 82/100 for self-testing and 76/100 for rereading; limitations: constructed data, no randomization, and short follow-up.
- Neither paper can be described as real academic evidence. The results are descriptive comparisons for teaching demonstration only.
Checks
Both PDFs were successfully parsed; each had one page and extracted body text. The CSV was checked for two literature rows, source filenames, sample sizes, results, and limitations. The review was checked for the explicit synthetic-paper label and the boundary that the materials are not real evidence. The deliverables are CSV and Markdown, so Word title placement, Excel column widths, and HTML 390px overflow checks are not applicable.
Limitations and Unverified Items
The materials provide no real participants, raw data, statistical tests, randomization details, or design basis for generalization. No statistical re-analysis or cross-check against real scholarly literature was performed. PDF text extraction was completed, but visual layout checking was not applicable because the deliverables are not PDF, Word, Excel, or HTML.
Download and inspect the outputs
The archive contains the files created during this run. Use these checks when adapting the example to your own data.
- Reconcile the reported results against the downloadable sample inputs.
- Open the generated files and check the contents relevant to your task.
- Review the limitations and any unverified steps in the report.