Task Modeling

Available in VPC

You can use the Task Modeling menu interface to create and delete Tasks.

Task Modeling interface

The Task Modeling interface includes the following components:

nclue-task-vpc_screen_ko

Component	Description
① Menu name	Current menu name.
② Basic features	Basic features provided in the Task Modeling menu. [Create a Task]: Click to create a Task [Features and pricing information]: Click to go to the NCLUE service overview page. [Refresh]: Click to refresh the Task list.
③ Delete	Click to delete a Task
④ Task list	List of created tasks Task ID: Unique ID automatically assigned when the Task is created. Task name: Name entered when creating the Task. Status: Current status of the Task. Registered: Task creation has been initiated. Pending: Waiting before proceeding to the next step. If there are many requests, the wait time may be longer. In progress: Task is being generated Available: Task generation completed and available for use Failed: Task generation failed. Ground truth data bucket: Name of the Object Storage bucket where the answer dataset file is stored. When you click the Train/Test detailed settings toggle button, the Train and Test answer dataset buckets are shown separately. Ground truth data file path: Path where the answer dataset file is stored in the Object Storage bucket. When you click the Train/Test detailed settings toggle button, the Train and Test answer dataset file paths are shown separately. Size: Size of the generated Task. System message: Message displayed when Task creation fails (see NCLUE Troubleshooting). Creation date: Date and time when Task creation was requested. Completed at: Date and time when Task creation was completed. Feature: Name of the feature used for task creation NCLUE model: Name of the NCLUE model used for feature creation. When you click the Train/Test detailed settings toggle button, the train and test features are shown separately. Row count: Number of rows in the feature set. AUROC: Area under the ROC curve; expresses Task Model performance as a value between 0 and 1.

Create a Task

You create a task by using a user feature together with an answer dataset. To create:

Note

Only one Task is created at a time, in the order that creation requests are received.
Task creation time may decrease or increase depending on available service resources.

In the VPC environment of the NAVER Cloud Platform console, navigate to Menu > Services > AI Services > NCLUE.
Click the Task Modeling menu.
Click [Create a Task].
When the Task creation popup appears, configure the new Task.
- Task name: Enter a unique name within 100 characters.
- Train/Test detailed settings: Click the toggle button to manually configure the data used for Train and Test.
  Note
  - You can use the dataset even if the labels are imbalanced, without any correction.
  - Train: Configure the training set used to train the model
  - Test: Configure the test set used to evaluate the trained model
  - If you do not configure these in detail, the NAVER Cloud Platform will split them automatically.
- Ground truth data file: Ground truth dataset file to use for Task creation
  Caution
  - Files containing personal information (resident registration numbers, passport numbers, driver's license numbers, credit card numbers, mobile phone numbers, emails) cannot be used.
  - There is no upper limit on the size of the answer dataset that can be used. However, Object Storage fees apply based on the storage size.
  Note
  - The minimum number of labels in the answer dataset is 100 samples each for labels 0 and 1.
  - You can use the dataset even if the labels are imbalanced, without any correction.
  - Feature: Select the feature you want to apply to the task model from the dropdown menu.
    - For details on creating features, Create a feature.
    - Click to refresh the dropdown.
  - Storage bucket: Enter the path where the Ground truth dataset is stored within the Object Storage bucket.
    - To use a new Object Storage bucket, click [Create an Object Storage]. However, after you create an Object Storage bucket, you must upload the ground truth dataset file to that bucket and click so that the bucket appears in the dropdown menu. For details, see Object Storage Bucket Creation.
  - File path: Enter the path where the ground truth dataset is stored in the Object Storage bucket.
    - The following special characters cannot be used in file paths:
      - &$@;:+,?*\{}^%`[]<>~#|"'
      - Characters with ASCII codes 0-31 and 128-255 are also not allowed.
      - Spaces
Click [Create].
- The Task appears in the list with the status Registered, and when Task creation is complete, the status changes to Available.

Note

Data for the created Task is stored in a space dedicated to the NCLUE service and cannot be viewed directly.

Caution

Check the version of the NCLUE model. Tasks created from features based on deprecated models cannot be used. We recommend using the latest models.

Delete a Task

To delete a Task:

Caution

Deleted Tasks cannot be recovered. Be sure to review the task before deleting it.
If the ground truth dataset used when creating the Task is no longer needed, delete it from Object Storage. If you do not delete it, Object Storage fees will continue to be charged based on the size of the ground truth dataset.

Note

If you select a Task in Pending or In progress status, [Delete] will be disabled.
You can select and delete multiple Tasks at once.

In the VPC environment of the NAVER Cloud Platform console, navigate to Menu > Services > AI Services > NCLUE.
Click the Task Modeling menu.
In the Task list, select the Task you want to delete, then click [Delete].
Review the information in the Task deletion popup, then click [Delete].
- The Task will be removed from the list.