This article explains how to use the data masking feature, including sensitivity classification for datasets, configuration of masking rules, and the results of applying masking.

Data masking refers to the process of applying masking rules to fields containing sensitive information to perform data masking. After enabling this feature, the relevant fields will be masked, transformed, or obfuscated according to the specified rules, thereby protecting sensitive data privacy in the workplace and fulfilling the company's compliance requirements.

Feature overview #

Benefits #

In the information age, requirements for data security are becoming increasingly stringent. Whether through government-issued laws and regulations or industry-specific guidelines, there are security requirements for personal information and other types of sensitive data. Data masking helps reduce the risk of sensitive data leakage and adds an extra layer of protection for private data. By introducing the data masking feature, the need for manual data masking in traditional business scenarios can be greatly reduced, improving data delivery efficiency and increasing satisfaction among stakeholders. At the same time, it enables data processors, analysts, and viewers to use data in accordance with the company's information security requirements, further strengthening enterprise information security.

Key advantages #

• Flexible masking rule configuration: Users can configure masking for specific fields based on data sensitivity, supporting various masking rules such as full masking, name masking, email masking, and contact information masking.
• Freely controllable scope of masking effectiveness: Users can set the scope of data masking according to specific business needs, configure the range of sensitive users, and achieve better business adaptability.

When saving a new dataset or editing an existing dataset, you can configure data masking rules.

Create a new dataset #

When saving a dataset, you need to select the "Confidentiality Level" option to determine whether the dataset contains sensitive information. If the dataset contains sensitive fields, set the level to "Sensitive"; otherwise, set it to "Not Sensitive".

• After selecting the "Sensitive" option, you must promptly add masking rules for sensitive fields under the "Data Masking" tab in the dataset details page; otherwise, the system will display an error message as shown below.
  
• If you select the "Not Sensitive" option, no further configuration is required. All data fields will not be masked, and the displayed field values will be the original values.

Edit a dataset #

1. Click "Data Preparation" - "Dataset", select a dataset, enter the dataset details page, and click "Data Masking".
   
2. Click "Add Masking Field", select the fields to be masked, and click "Confirm".
   
3. Add the corresponding masking rules to the fields that require masking. The system provides preset rule options for all sensitive fields, including full masking, name masking, email masking, contact information masking, and more. You can mask data according to preset rules or custom rules.
   
4. Set the scope of data masking, add users or user groups, and configure the rules for the scope of effect. The following two configuration methods are supported:
   • Applicable members are subject to rule control; users other than members are not subject to rule control.
   • Applicable members are not subject to rule control; users other than members are subject to rule control.
     
5. You can add multiple masking fields. After adding them, click "OK". After saving, you can view the masking results under the "Data Preview" tab.
   
6. Refresh to view the masked data, as shown below.
   

Dataset status #

Impact on associated datasets:

• When there is at least one masked sub-dataset, the logical dataset inherits the "sensitive and masked" status.
• The masked fields of the logical dataset are the union of masked fields from all sub-datasets.
• All adjustments to data masking rules in sub-datasets are synchronized to the logical dataset.

Data masking rules #

Preset masking rules #

Custom masking rules #

Scope of data masking effectiveness #

Effective scope #

The effective scope supports configuring a denylist or allowlist.

• Members to whom the rule applies are governed by the rules; users outside these members are not governed by the rules.
• Members to whom the rule applies are not governed by the rules; users outside these members are governed by the rules.

Designated members #

Users or department-level user groups can be added as designated members.

Effects of data masking application #

Visualization query results #

Description of masking status

When a dataset is sensitive, the following markers indicate different statuses.

Feature differences of masked datasets

• Supported: Sensitive fields can be used as dimensions, metrics, and filters; all chart and analysis functions are supported and operate normally.

• For specific unsupported features, refer to the following content:

  Scenario	Support status	Note
  Sensitive fields as dimensions	Supported	Supported for use as a dimension, but does not support functions such as "field value alias", "geographical role", or "create group".
  Sensitive fields as measures	Supported	Supports selecting aggregation methods for calculation. The system uses actual data to perform aggregation calculations and displays the results according to the masking rules configured for the field. Data format configuration is not supported.
  Sensitive fields as filter options	Partially supported	When sensitive dimension fields are used as filters, only exact filtering is supported, and the field's dimension values are also masked.
  Sensitive fields used for result filtering or detail filtering	Partially supported	Only scenarios where the aggregation method is "count" and the filter condition is "=" are supported; other aggregation methods or filter conditions such as greater than, less than, range, or not equal are currently not supported.
  OKR table	Not supported	The OKR table currently does not support the use of masked fields. If masked fields have already been used in the current analysis, switching to the OKR table is not supported.
  Trend analysis table	Not supported	The trend analysis table currently does not support the use of masked fields. If masked fields have already been used in the current analysis, switching to the trend analysis table is not supported; if the trend analysis table has already been selected, selecting masked fields is not supported.
  Calculated field contains masked field	Supported	The expression will inherit the rules for masked fields. If the expression is used as-is, it will be executed according to the data source field rules; if the expression is modified, it will be executed according to the rules for fully masked fields.

• When creating tables and charts, data in sensitive fields is masked.
  

• When a sensitive field is used as a dimension field, features such as "field value alias", "geographic role", and "create group" are not supported.
  

• When a sensitive field is used as a metric field, the "create group" feature is not supported.
  

• When a sensitive field is used as a chart metric, only the aggregation methods "Count", "Count distinct(fast)", " Count (distinct)", "Average","Maximum", "Minimum", "Variance", "Standard deviation"and "Quantile" are supported, and the "set data format" feature is not supported.
  

• When a sensitive dimension field is used as a filter, only "exact filtering" is supported, and the dimension values of the field are also masked.
  

• Calculated fields: When a field expression contains a sensitive field, the sensitive field is displayed in red, and the expression inherits the rules of the masked field.
  

  • When the expression is not processed at all, it is executed according to the data source field rules.
  • When the expression is processed, it is executed according to the fully masked field rules.

Dashboard effects #

• When the global filter contains sensitive fields, the values of sensitive dimensions are masked.
  
• For data download and dashboard export functions, all data in sensitive fields is masked. The left side of the figure below shows downloaded data, and the right side shows the dashboard.
  
• The results displayed in dashboard subscription, monitoring, and embedding functions are all masked.

Visual modeling effects #

The dataset output from visual modeling does not inherit the masking field controls of the input dataset. After outputting as a new dataset, the classification status is 'unclassified', and the creator must reconfigure the masking fields. The configuration operations are consistent with the content described earlier in this document.