Scalable Conflict of Interest management and analysis system
Find a file
Luka Dekanozishvili 9c26bc4f4b init commit
2026-06-02 19:29:51 +02:00
misc init commit 2026-06-02 19:29:51 +02:00
notes init commit 2026-06-02 19:29:51 +02:00
src init commit 2026-06-02 19:29:51 +02:00
.gitignore init commit 2026-06-02 19:29:51 +02:00
pyproject.toml init commit 2026-06-02 19:29:51 +02:00
README.md init commit 2026-06-02 19:29:51 +02:00

Scalable Conflict of Interest management and analysis system

WIP documentation

Installation

Install uv.

Usage

Conflict of Interest detection (main feature)

WIP

Qualify

The "qualify" feature, given an input list of users, outputs a list of qualified reviewers that have a certain amount of combined papers in the specified venues. These values can be seen in src/constants.py.

uv run src/main.py qualify --xml data/dblp.xml --output output/qualified-reviewers.csv --users input/Users.csv

Optionally add the --reverse flag to output only the unqualified reviewers.

See for detailed flags:

uv run src/main.py qualify -h

The output includes these additional fields:

total_qualifying_papers: Sum of the papers published by the author in e.g. SIGMOD, VLDB, ICDE, EDBT, CIDR

total_publication_count: Sum of all papers in DBLP by the author

disambiguation_level: With values:

  • certain: The matching from the input user to the DBLP entity was done by a URL/ID
  • doubt: The matching was done using heuristics (organization/country/author name)
  • failed: All of the people with the same name were merged into one entity. These results can be wildly inaccurate

Notes

There's a script ./misc/input-statistics.py that calculates how "full" the data is and what URLs are missing.