Overview
SRVC is an open-source document review system. It is designed to make common review tasks fast and easy. Here is an overview of how to create a review project with SRVC.
Create a project
Every SRVC project has an sr.yaml configuration which
specifies labels, review flows, and other parameters.
See the sr.yaml
from the srvc-template repository:
reviewer: mailto:user@example.com
labels:
category:
question: Category
type: categorical
categories:
- A
- B
- C
include:
json-schema: boolean
question: Include?
required: true
flows:
pubmed-search:
steps:
- uses: github:insilica/srvc-pubmed-search
query: angry bees
label:
steps:
- run-embedded: generator sink.jsonl
- run-embedded: remove-reviewed
- run-embedded: label-web
labels: [include, category]
port: 5005
To follow along, create a directory for your new project, and place this example sr.yaml
file in that directory.
Note: This example requires Nix to be installed.
Import documents
With the sr.yaml
file given above, documents can be imported from PubMed with the command:
sr flow pubmed-search
As the search step runs, you can see documents being appended to a file named sink.jsonl
.
This step only needs to be run once, but can be repeated without harm. Any new or changed documents will simply be added to the existing documents.
Review documents
Reviewing documents involves assigning answers for the “include” and “category” labels. Get started with the command:
sr flow label
You should see the text Listening on http://127.0.0.1:5005
.
Visiting that address in a browser will load an interface that shows a document and allows submitting answers for that document.
After you submit an answer, you should see that it has been appended to sink.jsonl
.
Collaborate
A remote sink such as Sysrev can be used to keep data in sync with other reviewers.
Another way to collaborate with other reviewers is to use a git repository for your SRVC project directory. Reviewers can merge their work by merging their sink files. If there are any git conflicts, simply accept all additions from both reviewers. The SRVC data model allows for seamless merging.