Action: Create Projects

Automate creating projects using files directly from your object storage

This Action can be accessed from the sidebar which is called Automation.

External object storage is required for this feature to work. Datasaur will access the files from a specified path in your bucket for the automations.

Project template is also required. Datasaur uses the template to configure what kind of projects that will be created.

⚠️ Please note that currently we are still in the progress of supporting Bounding Box labeling with Action.

Creating and/or Edit an Action

  • To create, click the New Action button on the page.

  • To edit, click on the triple dot menu at the top right corner of an Action card, then click Edit.

See the more detailed explanation for some of the fields.

Step 1

All fields are required to be filled.

  1. External object storage. It must be chosen from an existing one. You have to integrate your object storage first before setting up the action. If the Action is run, it will fetch the data from this bucket.

  2. Project template. Same as above, you have to create a project template first before setting up the action. The projects that will be created through the Action will follow the exact configuration of the project template.

  3. Input path. When running the Action, Datasaur will iterate files from this path. Each folder represents a project and each file inside it will be treated as the documents.

    1. Please note that this path is from the root of the bucket. So, if you want to process S3://test-external-object-storage/input, you will only need to fill input as the value. The actual bucket information is already being configured previously via External Object Storage that you've chosen on the other field.

  4. Result path. After successfully creating a project, Datasaur will move the files to the result folder so that the next time the same Action is run again, the files won't be processed twice. It's essential that you don't move any files inside this result path because it would make Datasaur cannot access the files appropriately. If file conversion process is required (e.g. converting docx to PDF), the converted files will be also available in this folder.

  5. Tags (optional). You can add custom tags to each project that is created through Action by filling this field.

Step 2

  1. Assignees. Select a pool of labelers and reviewers for the Action that will be distributed based on your preferences. There has to be at least one labeler and one reviewer.

  2. Number of labelers per project. Each time a project is created with this Action, this field will determine how many labelers (from all the selected labelers above) that will be assigned to the project. If it's lower than the total number of labelers, the assignment would be distributed using a round-robin algorithm.

  3. Number of reviewers per project. Exactly like the above but for reviewers.

Step 3

  1. Number of labelers per document. Essentially, it's the same as above. But this assignment is specific for each document inside the project.

  2. Conflict resolution. Please click here to get a more detailed explanation.

Step 4

It's basically a recap of what the settings look like coming all together. There is also a preview using dummy data of how the projects are going to be configured, including the assignment illustration when the Action runs.

Running the Action and See the Result

From the App

  1. Go to Actions.

  2. To view the activities of an Action, click on View Run.

    1. Each time an Action is run, it will be represented as one Action Run (one row) which contains information related to the automation process. You can also see the previous runs.

    2. Note that in one Action Run, it could create more than one project, which will be covered in the View Details.

    3. If the View Details is not available, then there is no new data found on the bucket. Hence, the Action cannot create any projects since there aren't any folders with files that can be processed.

  3. Action Run Detail will show the information regarding each attempt when creating a project (one row represents one attempt).

Through API

To do an API call, start from here.

Last updated