Private by design
Compare permitted submission metadata with your research profile using an embedding model that runs entirely in this browser.
Step 1
Each item is embedded separately so matches can point back to its strongest influence.
Step 2
CSV and JSON stay in memory unless you explicitly enable local persistence.
Expected fields: paper ID, title, and ideally abstract, topics, track, preference.
Step 3
Checking which backend this browser will use…
Default scientific-paper retrieval model. Downloads about 441 MB from Hugging Face, then embeds all paper text locally.
Step 4
Read the paper metadata and make your own judgment before recording a preference.
Select a cluster of your publications and map the closest submissions.
No papers imported yet. Set an interested target if your chair asked for a quota.
Interested papers boost similar candidates. Marking a paper not interested can optionally downrank close variants or add negative phrases. Conference topics are not used for scoring unless you add them as negative phrases.
No rankings yet.
Publication cluster
Choose publications that represent one subject. The map and Top X list compare imported submissions only with that selected cluster.
No similarity results yet.
Local persistence
Enabled by default. Profile text, paper metadata, scores, statuses, and notes are stored only in this browser’s IndexedDB. Turn it off to remove the saved copy.
This session will be saved locally.
The JSON backup includes your expertise profile, imported papers, rankings, review decisions, notes, interested target, and export settings. Importing replaces the current session.
Export
Use blank or an integer from −20 to 20. Conflict-marked papers are omitted; declare conflicts through the official conference process.
Technical overview
The ranking is a local semantic comparison, not an assessment of reviewer fitness.
Your weighted expertise items and each paper’s title and abstract are normalized and split into overlapping chunks when they exceed the model’s practical context size. Imported conference topics remain available only as display/filter metadata.
SPECTER2 proximity is the default scientific-paper retrieval model and downloads on demand. The quantized all-MiniLM-L6-v2 model is served with the app as a faster offline option. EmbeddingGemma 300M and BGE-M3 can also be downloaded on demand. SPECTER2, MiniLM, and BGE-M3 use WebGPU when available, with a WebAssembly fallback. EmbeddingGemma uses FP32 weights on WebGPU and is unavailable when WebGPU is not supported.
Chunk vectors are mean-pooled and normalized. Cosine similarity compares each paper with every profile item. The final score aggregates up to three near-best positive matches with decaying weight; unrelated expertise areas do not lower a strong topical match.
The strongest matching profile items explain each base score. Interested decisions add a small similarity boost. Not-interested papers can optionally downrank close variants, and explicit negative phrases apply a stronger topic-level penalty. Similarity never becomes a bid automatically, and conflicts remain part of the official conference process.