Small decisions. Interesting possibilities.Submit contentSubmit

Jev Reviewer

Introducing Jev Reviewer, a tool designed for systematic reviewers to quickly find and extract the information from research articles. App: GitHub:

Added to Jevfast

How it uses Jev

The browser reads study files into located text blocks and asks Jev to select candidate line IDs for each extraction question, then uses typed yes/no judgments to verify selected lines. Code copies matching source text verbatim with its file and location; Jev does not write the extracted answer.

What Jev decides

Illustrative Jev requestExample answers · not a recorded Jev response · Source ↗
Question 1 · verify line
YOUR APP
INSTRUCTION

Does line A042 of lines itself state information that answers request?

STATE

A segmented article passage, extraction request and a candidate line A042.

JEV · NOUL
YesNo

App workflow

  1. Read and segment study files

    Browser readers convert supported documents, tables, slides, and saved pages into ordered blocks with file-specific location IDs.

  2. Screen candidate passages

    For each chunk, Jev ranks candidate lines against the project's extraction questions and can indicate that no line fits.

  3. Verify and quote

    A second Jev pass checks candidate lines; code copies qualifying text verbatim, merges adjacent lines where applicable, and attaches source locations.

  4. Human-check and export

    The reviewer checks or edits answers, compares studies in an extraction table, and exports checked values and supporting quotations.

Why it is interesting

Its two-pass design distinguishes relative candidate retrieval from absolute answer verification, and preserves provenance so reviewers can check quotations in context instead of trusting generated paraphrases.