Two steps on one page: a matching scores every submission against every pool member, then the assignment is solved from those scores under your constraints.

1. 1 Taxonomy (uses an LLM) builds a topic taxonomy for your venue and scores a pair by how close the paper's topics are to the reviewer's. It is the matching you can read: topics are named and every match shows the publications behind it.
2. 2 Expertise Prototype (no LLM) scores a paper against prototypes of each reviewer's publication history. Nothing leaves the server.
3. 3 Your own scores, from bids or another system, if you only want the assignment solved.
4. 4 Your data: what the Taxonomy matching sends to the LLM provider, and that a venue is deleted after 30 days.
5. 5 Options: reviewers per paper and papers per reviewer as ranges, a minimum match score, country diversity, no co-authors, and the same for area chairs.
6. 6 Conflicts of interest are always applied: a conflicted pair is never assigned. Manage opens the conflicts page (below).
7. 7 Run a matching with the button on its card (it then solves too), or re-solve here after changing only the options. Each matching's scores are kept, so you can switch between them.

1. 8 The result in numbers: pairs assigned, papers below the minimum, the lowest assigned score, the conflicts kept apart, and the spread of loads.
2. 9 Insufficient coverage names the papers with too few reviewers and the topics short of reviewers. When the pool is too small, as here, the places it has are spread evenly: no paper is two reviewers short while another is full.
3. 10 One row per submission: its best match in the pool, its reviewers and its area chair, each with their match on a 0–100 scale (100 is the venue's closest pair).
4. 11 Why this reviewer: their three evidence publications closest to the paper, each with its own match and the topic it sits in.
5. 12 Hide the publications to see more papers at once. Click a title to open the paper.

1. 1 The paper's topics in the taxonomy, with the share of the paper in each.
2. 2 The reasons for each assigned reviewer.
3. 3 Lock a pair to keep it when you solve again; Remove one to keep it out.
4. 4 COI declares a conflict of interest for the pair and takes it out of the assignment.
5. 5 Add & lock any other candidate; the list is ranked by match and shows each person's load.

1. 1 Upload the conflicts your submission system collected: one row per submission and pool member, the member by id or by email address.
2. 2 The template shows the columns.
3. 3 Find authorship conflicts from the data already here: pool members who are authors of a submission, or have written with one.
4. 4 Add one by hand.
5. 5 The list, by where each conflict came from; a new upload replaces the uploaded ones only.