Data

  • Edit Times:3
  • Recent Updates:Charis.Dong on 2026-06-17
  • Description

    Function Description

    The Data module is the core carrier in Data Agent to provide analysis materials for agents. All business data available for agents to query and analyze is uniformly managed and configured here. The Data module supports capabilities such as scheduled refresh, folder management by group, and semantic configuration, ensuring data accuracy and business consistency during agent-based data analysis.

    Note: Currently, only FR/BI analysis subjects and dashboards can be connected. Excel files and third-party databases will be able to be connected in the future.

    Application Scenario

    • Agent data configuration: Bind one or more data assets to an agent, enabling the agent to perform tasks, such as conversational analysis and report generation, based on these data assets.

    • Unified data management: Organize data sources from different business lines into different folders.

    • Scheduled data refresh: Set the automatic loading and refresh frequency to ensure data timeliness.

    • Unified semantic management: Configure business terms (synonyms) for fields to improve natural language recognition accuracy.

    • Data asset connection expansion: Add analysis subjects or dashboards from FR/BI data assets.

    Effect Demonstration

    Taking how to add the data asset Chain Operation Analysis from an FR/BI analysis subject as an example, the following figure shows the effect.

    数据 图1.png

    Data Directory

    Data Viewing and Filtering

    Choose Management Backend > Data. The left data directory tree displays all data assets. You can quickly locate data using the search box at the top, as shown in the following figure.

    数据 图2.png

    Click the filter icon to filter data assets by source category. The following lists currently-supported categories:

    • All

    • FR/BI Analysis Subject

    • FR/BI Dashboard

    数据 图3.png

    Directory Management

    You can perform the following operations on folders/data assets in the directory:

    (1) Rename: You can rename existing folders/data assets for easier business scenario identification.

    (2) Move to: You can move folders/data assets to a specified folder, enabling business scenarios to be managed by group and the directory structure to be adjusted.

    (3) Delete: You can delete data assets/folders that are no longer in use.

    数据 图4.png

    Note:

    (1) As the root directory, the All Data folder cannot be renamed, deleted, or moved.

    (2) Folder/Data asset deletion will affect related agents and skills, and is irreversible. Delete them with caution.

    Data Asset Adding

    Analysis Subject Adding

    Choose + > Analysis Subject to open the Add Analysis Subject dialog box. In the left directory tree, select the analysis subject you want to import. You can click the analysis subject name to preview its field structure on the right. After renaming the data asset, click OK to complete the analysis subject import.

    数据 图5.png

    Dashboard Adding

    Choose + > Dashboard to open the Add Dashboard dialog box. In the left directory tree, select the corresponding dashboard. You can click the dashboard name to preview its content on the right. After renaming the dashboard, click OK to complete the dashboard import.

    数据 图6.png

    Data Preview and Configuration

    Data Preview

    Click any data asset in the left data directory to preview real-time data on the right.

    (1) Analysis subject: You can preview data details and field information by page and search for fields.

    数据 图7.png

    (2) Dashboard: You can preview the complete dashboard effect. Operations, such as Save Dashboard As, Export, Adaptation, Performance Analysis, Indicator Info, and Original Query Condition, are supported for dashboards.

    数据 图8.png

    Field Information Viewing

    In the right preview area, you can switch to the field tab page.

    (1) View the field type, field name, and field alias of each field.

    (2) Set the data format (automatic, number, or percentage) and metric type for number fields.

    • Additive Metric: Metric whose values can be directly summed  across all dimensions

    • Semi-Additive Metric: Metric whose values cannot be directly summed by some dimensions, but can be summed using specific      aggregation methods

    • Non-Additive Metric: Metric whose values cannot be directly summed by any dimension

    数据 图9.png

    Semantic Information Configuration

    Semantic information is used to unify the business terms and synonyms for data fields, improving the agent intent recognition accuracy.

    1. Click the Semantic Info button on the right to enter the semantic configuration page.

    数据 图10.png

    2. (Optional) Add/import/exporting business terms:

    • Add: Select the object type, and enter the standard object and business terms (synonyms). For example, map Sales with Revenue and Turnover.

    • Import/Export: Manage business terms in batches for easy  cross-scenario reuse.

    数据 图11.gif

    3. After the configuration, click Detect and Save. The agent will automatically recognize these synonyms.

    数据 图12.png

    Note: The semantic information configuration takes effect only for the current data asset. Different semantic information needs to be maintained independently for different data assets.

    Data Refresh Configuration

    To ensure the accuracy of data used by agents, data can be preloaded and refreshed periodically.

    1. Click the  icon to open the data refresh configuration page.

    2. View the last refresh time. You can click Refresh Now to manually synchronize data.

    3. Configure the automatic refresh frequency.

    4. (Optional) Configure the start time. If it is not configured, data will be refreshed at the top of every hour by default.

    数据 图13.png

    Note: You are advised to set the data refresh frequency based on business real-time requirements to prevent excessive refresh from affecting performance.

    Attachment List


    Theme: Data Access
    • Helpful
    • Not helpful
    • Only read