Workflow:ThreeSR Awesome Inference Time Scaling Automated Paper Addition
| Knowledge Sources | |
|---|---|
| Domains | Academic_Curation, Automation |
| Last Updated | 2026-02-14 01:30 GMT |
Overview
End-to-end process for automatically adding academic papers to the curated inference-time scaling paper list using the Semantic Scholar API.
Description
This workflow automates the addition of new papers to the repository's README.md paper list. It queries the Semantic Scholar API to retrieve paper metadata (title, authors, arXiv ID, abstract, publication date, venue), formats the metadata into the repository's standard markdown template, merges the new entry into the existing list while preserving reverse-chronological sort order, deduplicates against existing entries, and commits the result to git.
The output is an updated README.md with the new paper entry inserted in the correct chronological position, followed by an automatic git commit and push.
Usage
Execute this workflow when you want to add a new paper to the curated list and you know the paper's title or a distinctive keyword from it. The paper must be indexed in Semantic Scholar. This is the preferred method for adding papers because it ensures consistent formatting and correct chronological placement.
Execution Steps
Step 1: Configure Search Query
Parse command-line arguments to obtain the paper name or search keyword. The script accepts a --paper_name argument that will be used as the query string against the Semantic Scholar API. If no argument is provided, the default query is "Inference-Time Scaling".
Key considerations:
- Use the exact paper title for precise results
- Partial titles or keywords will return the closest match
- The search limit defaults to 1 result (the most relevant match)
Step 2: Search Semantic Scholar
Query the Semantic Scholar Graph API search endpoint with the configured search term. The API returns structured paper metadata including title, authors, venue, year, publication date, fields of study, and a canonical URL. Results are sorted by year to prioritize recent publications.
Key considerations:
- The API endpoint is api.semanticscholar.org/graph/v1/paper/search
- Rate limits apply; avoid rapid successive calls
- An alternative date-range search function exists but is not used in the default flow
Step 3: Fetch Extended Paper Details
For each paper returned by the search, fetch additional details from the Semantic Scholar paper detail endpoint. This retrieves the arXiv ID and full abstract, which are not included in the initial search response.
Key considerations:
- Uses the v1 paper detail endpoint with the paperId
- The arXiv ID is required to construct proper arXiv links
- If no arXiv ID is found, links will contain "N/A" placeholders
Step 4: Format Paper Entry
Convert the retrieved paper metadata into the repository's standard markdown template. The template includes the paper title as a link to the arXiv abstract page, a direct PDF link, author list, publication date, venue/publisher, and a collapsible abstract section.
Template structure:
- Title line with arXiv abstract link (prefixed with emoji marker for parsing)
- arXiv PDF direct link
- Comma-separated author list
- Publication date in YYYY-MM-DD format
- Venue name (defaults to "arXiv.org" if empty)
- Abstract in a collapsible HTML details element
Step 5: Merge and Sort Into README
Read the existing README.md, locate the paper list section by its heading, parse all existing paper entries, check for duplicates by title, merge the new formatted entry with the existing entries, sort all entries by publication date in descending order (newest first), and write the updated content back to README.md.
Key considerations:
- Entries are delimited by the emoji marker at the start of each title line
- Duplicate detection compares exact titles; slight variations will not be caught
- Papers with unparseable dates are sorted to the end of the list
- The section boundary is detected by the next markdown heading
Step 6: Commit and Push Changes
Stage all modified files, create a git commit with a default message, and push to the origin/master remote. This finalizes the paper addition and makes it available to all users of the repository.
Key considerations:
- The commit message defaults to "Update"
- All files are staged (not just README.md)
- Push targets origin/master specifically