When using Intelligent Data Insight (DataWind) together with the Agent, you can use charts saved in DataWind as data sources for the intelligent analysis Agent. By configuring question examples and other settings, you can improve the efficiency and accuracy of chart-related data queries. This article provides detailed instructions for creating data sources based on charts.

Agent data source configuration supports two types of data sources:

• Add an existing dataset as a data source for the Agent.
• Add a visualization chart created in Intelligent Data Insight (DataWind) as a data source for the Agent. This method is supported starting from version 3.2.0, and you need to use DataWind together with the Agent.

This article provides detailed instructions for creating data sources based on charts in DataWind; for details on configuring Agent data sources based on existing datasets, see Dataset Configuration.

You must first create the corresponding visualization chart in Intelligent Data Insight (DataWind). For details, see DataWind Help Documentation: Overview of Visualization Queries.

After entering the Agent management configuration page, click "Data source configuration" in the left navigation bar. Then, after entering the data source configuration page, click the "Chart" tab to access the chart configuration page.

Add method 1: Manually select a chart #

1. Click "Add chart" in the upper right corner of the page. In the pop-up window, select the chart to be added, and then click "Next" when finished.
   

   1. In the left section of the pop-up window, you can switch the project and dashboard directory where the chart is located, or quickly locate the dashboard containing the chart to be added by searching.
   2. In the right section of the pop-up window, you can select the charts in the dashboard to be added. Multiple charts can be selected and added at the same time.

2. On the "Configure rules" page, configure the chart name, question examples, and chart description.
   

   Parameters	Configuration instructions
   Chart name	By default, the chart name in your dashboard is used as the name of the intelligent agent data source chart. You can also manually modify the chart name to a more understandable name to facilitate subsequent chart identification and improve the model's understanding of the chart.
   Question examples	Configure the most relevant question examples for the current chart. After configuration, when users use the intelligent analysis Agent to query data, if the data query is similar to the question examples configured here, the intelligent agent will return the current chart as the answer, improving the efficiency and accuracy of data queries.
   Chart description	Add detailed content such as chart definitions and data explanations to help the intelligent agent understand the chart more accurately.

3. After completing the configuration, click "Add" to add the chart as a data source for the agent.

Method 2: Batch addition using a table template #

1. Click "Batch import" in the upper right corner of the page. In the pop-up window, download the import template as prompted by the interface.
   
   When importing, the supported file formats are csv, xlsx, and xls. Each file must not exceed 100 MB.

2. Configure the parameters of the charts to be added in the downloaded import template.

   Parameters	Configuration instructions
   Chart embed URL	Configure the chart's embed URL. You can find the corresponding chart on the DataWind dashboard page, click Embed Chart, and obtain the embed URL for the chart.
   Chart ID	The chart ID can be obtained from the chart embed URL. The typical format of the chart embed URL is: https://xxxx/dashboard/35**7/183***3?appid=xxxxx, where the second ID after 'dashboard' is the chart ID, which in the above example is 183***3. You can also find the corresponding chart ID on the DataWind visualization query page. For details, see rid（reportid）.
   Chart name	By default, the chart name in your dashboard is used as the name of the intelligent agent data source chart. You can also manually modify the chart name to a more understandable name to facilitate subsequent chart identification and improve the model's understanding of the chart.
   Chart description	Add detailed content such as chart definitions and data explanations to help the intelligent agent understand the chart more accurately.
   Common question examples	Configure the most relevant question examples for the current chart. After configuration, when subsequent users use the intelligent agent to query data, if their data query questions are similar to the question examples configured here, the intelligent agent will return the current chart as the answer, improving the efficiency and accuracy of data queries.

3. After completing the configuration, add the configured import template as prompted by the interface pop-up to batch import charts.

After completing the chart data source configuration, you can view the list of added charts on the page.

• After adding the chart data source, the agent will automatically extract the configured chart name, description, example questions, and other content, and enrich them semantically using the original chart information (such as table headers, original chart name, description, and so on). You can view the semantic enrichment status of each chart data source in the chart data source list.
• You can view information such as the source dashboard and dataset for the corresponding chart.
• Click the edit or delete button next to the chart to edit or delete the chart data source.

After completing the configuration of the chart data source, when subsequently using intelligent Q&A for data queries, if the query concerns data related to the chart or the question is related to the example questions of the chart, the intelligent analysis Agent will return the corresponding chart. By explicitly linking the chart, query accuracy can be further improved, user experience enhanced, data sources expanded, and configuration efficiency increased.

• You can directly filter, sort, and perform other operations on the chart in the results returned by intelligent Q&A.
• You can hover your mouse over the dataset name below the chart to view the details of the dataset for the current chart.
• You can click the jump button in the upper left corner of the chart to quickly navigate to the chart editing page and quickly view more chart details.