A queue's configuration defines its capabilities, layout and properties in the underlying data model. This is done through the New and Edit Queue functionality, which exposes the Queue Builder.
If we do Home > New Queue, we navigate to an empty Queue Builder screen, which lets us build a queue from scratch:
From here, we can start constructing our queue. A queue is composed by leveraging the components on the left hand-side panel, which we call Blocks. Blocks can be drag-and-dropped into the Queue canvas to create a UI of your preference.
Every time you drop a new block into the canvas, the block's settings dialog will pop up, as illustrated above. Each block type will have generic and type-specific parameters.
The following settings exist for all types:
Label. Identifies the block towards end users. Shown at the top of the block.
Identifier. The block's identifier, used for technical integrations (ie API, Webhooks, CSV...)
These other settings might appear for several, but not all, types:
Placeholder. A sample value used to exemplify how the queue will look like in the builder view.
Read-only. Makes the block's value immutable inside the queue. In these cases the value can only be set once per task when it is created.
Required. Disallows task completion until the block's value is non-empty.
Use Placeholder. Renders the Placeholder as a fallback when no value has been supplied for a particular block in the task.
By hovering over the block, you can see its settings and delete buttons, as well as move and resize handles to adjust each block in the layout:
Below are all the Block types currently supported in Human Lambdas.
Displays a string of text.
Displays numeric data, no reformatting applied.
Displays an image, as long as its underlying value points to a valid static image URL.
Plays an audio, as long as its underlying value points to a valid static audio URL.
Displays a video, as long as its underlying value points to a valid static video URL.
Binary question, with Yes/No options, represented as
Single choice question. Similar to how Blocks work, each option has a Label and Identifier for end users and technical integrations respectively.
Multiple choice question. Similar to how Blocks work, each option has a Label and Identifier for end users and technical integrations respectively.
Displays a sequence of questions. The question types supported are:
The questions will be presented sequentially from top to bottom. There is limited support for logic jumps — useful to build conditional paths — under the following types:
Displays an email address, and includes validation on input.
Displays a URL link, and includes validation on input.
Displays a website through an iframe.
Renders a PDF given a valid .pdf URL. Uses the browser's PDF viewer.
For Named Entity Recognition tasks, it allows the user to annotate segments of a text and map them to various predefined categories.
A Rich Text editor with support for Markdown and HTML text formatting.
For Object Detection annotation tasks, it allows the user to draw categorized bounding boxes over an image file, supplied by its URL.
Represents a sequence of strings. Useful for cases where the task might require processing an indeterminate number of text items.
We're actively working more block types to the catalog. If you have a specific need that's not supported by the current blocks and their existing functionality, drop us a line at [email protected] and we'll be more than happy to take a look.