Tags
Overview
Tags use classification rules, scoring criteria, or extraction fields to turn information in posts, videos, and comments into structured results for filtering and analysis.
Open Tag from the top-level navigation to manage tag tasks and view their rules and results. Tag results can be used in Social Media Content filters and Visual Analysis.
| Tag type | Main use | Example |
|---|---|---|
| Classification tag | Assign content to predefined categories | Identify discussions about hero adjustments, ranked play, or skin events |
| Scoring tag | Score content against consistent criteria | Quantify player satisfaction with a new season |
| Extraction tag | Extract specified information from content | Extract hero names, event names, and other fields from discussions |
Tag results include category values, scores, and extracted fields. Tags process only content already received by the platform; they do not change the data collection scope.
Tag list page
The tag list shows the tags in the current space, their data ranges, types, update settings, and task statuses. It supports search by name and list filters.
Click a tag name to view its rules and tag overview. Depending on your account permissions, you can also create, edit, copy, or delete tags, or run tag tasks manually for a specified time range.
Create a tag
Click Create Tag in the upper-right corner of the tag list, then configure the basic information, tag rules, and update settings.
Basic information
- Tag Name: Describe the processing goal, for example, “New Season Player Satisfaction” or “Hero Names Discussed.” The name cannot be changed directly after creation.
- Data Range: Select the channels and content types to process. Posts, videos, and comments are currently supported. You can also choose whether to automatically include channels added to the space later and specify which content types to process for those channels.
- Tag Type: Choose a classification, scoring, or extraction tag, shown in the interface as Category, Rating, and Extraction. The type determines how rules are written and the format of the results. It cannot be switched directly after creation.
Classification tags
Classification tags assign content to predefined categories, such as discussion subjects, player attitudes, or the gameplay, heroes, and events mentioned.
Enter a name and recognition rules for each category. The name becomes the resulting tag value. The rules should explain how to identify the category and distinguish it from other categories.
For example, a “New Season Player Attitude” tag could define Positive, Neutral, and Negative categories, with criteria for agreement, factual discussion, and clear dissatisfaction.
Additional settings include:
- Task Requirements: Limit which content the task processes, for example, discussions related to the current season only.
- Matches: Set the maximum number of categories that one content item can match. If multiple matches are allowed, the result can contain multiple category values.
- Default Value if No Match: Use a default category value, or use Null to indicate that there is no match.
- Sample Data: Provide input content and expected output to clarify the classification criteria.
The following are input and output examples for “New Season Player Attitude”:
Input: The matchmaking experience has improved noticeably this season.
Output: ['Positive']
Input: New ranked rules have been added this season.
Output: ['Neutral']
Scoring and extraction tags
Scoring tags
Scoring tags score content against consistent criteria, for example, to quantify satisfaction with a new season based on player opinions.
Configure the minimum value, maximum value, number format, and scoring criteria. Number formats include integers, one decimal place, and two decimal places. The criteria should explain what qualifies for low, high, and intermediate scores.
You can also provide constraints, a default value when there is no match, and sample data. If a valid score cannot be identified, the result can remain empty or use a default number.
Extraction tags
Extraction tags extract specified information from content. A tag can contain multiple fields, each with a name, data type, and extraction rules.
Supported types are text, numeric, time, Boolean, and list. For example, “Hero Names Discussed” can use a list to retain multiple hero names from one content item, while “Mentions Ranked Play Issues” can use a Boolean value.
The extraction rules should explain what to identify, how to handle equivalent expressions, and how to handle missing information. Depending on the field type, you can also set a length limit or default value for unmatched content, and add constraints and sample data to clarify the task.
Update settings
All three tag types support automatic updates. When enabled, the system processes content from the previous period according to the schedule. You can currently schedule a daily run at a specified time; the page displays the execution time zone as UTC+8. When automatic updates are off, you can run the task manually as needed.
Schedule automatic updates after data collection and ingestion. To process historical content, select the corresponding time range and run the task manually. Creating a tag does not automatically tag historical content.
Tag details page
Click a tag name in the list to view its data range, content types, and update settings, and switch between two tabs:
- Rule Details: View classification rules, scoring criteria, or extraction fields, along with default values, task requirements or constraints, and sample data.
- Tag Overview: Select channels and a time range to view completed tag results. Classification tags show category distributions, scoring tags show score distributions, and extraction tags show field results.
Editing, copying, and manual execution on the details page are subject to account permissions.
Run a tag manually
To process historical content or content from a specified period, open Run Now from the tag list, details page, or creation confirmation.
- Select a start date and end date to define the publish time range of the content to process.
- Confirm the range and task rules, then click Run.
- The task runs in the background, so you can leave the page. View its status in the tag list and its results in the tag overview.
Tag tasks consume LLM tokens. Processing time depends on the amount of content and the complexity of the rules. After submitting a task, check its status in the tag list instead of repeatedly submitting the same range.
After processing, use tag results to filter content in Content Center, or combine them with other fields in Visual Analysis.

