All Products
Search

This Product

    Quick BI:Question-based Knowledge Management

    Document Center

    Quick BI:Question-based Knowledge Management

    Last Updated:Jun 23, 2026

    The knowledge base lets you define enterprise-specific knowledge and terminology. The model learns this information to improve the accuracy of data retrieval and analysis.

    Important

    This feature is available only in the China (Hong Kong) , Malaysia (Kuala Lumpur) and Singapore regions. It is being progressively rolled out to other regions.

    Access

    To access the Question-based Knowledge Management page, follow the steps shown in the image.

    image

    You can configure rules for business logic, regular expression match, dataset selection, and AI-powered table selection.

    Business logic

    On the Q-Data knowledge management page, click Business logic to configure the following settings.

    Add business logic

    image

    1. In the upper-right corner of the Q-Data knowledge management - Business logic page, click Add business logic.

    2. Configure the business definition, data interpretation, synonym, effective scope, and application method.

      • Business definition: Define a common concept within your enterprise (for example, "sales progress"). Maximum length: 100 characters.

      • Data interpretation: Describe the business definition in detail and link it to data metrics to help the model understand it. Maximum length: 300 characters.

      • Synonym: Define alternative names for the business term used within your organization, so the model can recognize different phrasings of the same question.

      • Effective scope: Select All enterprise data or Apply to specific data only. When a user asks a question within the specified scope, the model references this knowledge during inference.

        • All enterprise data: Applies to all Q-Data resources in your enterprise.

        • Apply to specific data only: Select the Q-Data resources to which the rule applies. You can search by Q-Data resource name.

          image

      • In the Application method section, choose whether to Enable forced rewriting.

        Note

        If you enable forced rewriting, any part of a user's question that matches the business definition or its synonym is automatically replaced with the data interpretation. Use this option with caution.

    3. Click Save.

      To add another entry, click Save and continue adding.

    Batch import

    1. Click Batch import to open the Batch import business logic dialog box.

      image

    2. Click Enterprise knowledge base - business logic template to download the template.

      image

    3. Click to upload or drag and drop a file.image

      Note

      Only .xls, .xlsx, and .csv files are supported. The file size must be 5 MB or less.

    4. After uploading the file, click OK.

      image

    5. After a successful upload, the imported entries appear in the knowledge base.

    Batch export

    Click the image icon to batch export all business logic rules.

    image

    Regular expression match

    On the Q-Data knowledge management page, click Regular expression match and configure the following settings.

    Add regular expression match

    image

    1. In the upper-right corner of the Q-Data knowledge management - Regular expression match page, click Add regular expression match.

    2. Add a business definition, regular expression, and data interpretation, and then set the effective scope and application method.

      • Business definition: A name for the regular expression. It is not used for question matching. Maximum length: 100 characters.

      • Regular expression: Used to identify patterns in user questions and execute actions based on the application method. Use Python-style syntax. Maximum length: 100 characters.

        You can enter text to match to test the expression and view the match result.

      • Data interpretation: Provide a detailed explanation of the business definition and link it to data metrics to help the model understand the pattern. Maximum length: 300 characters.

      • Effective scope: Select All enterprise data or Apply to specific data only. When a user asks a question within the specified scope, the model references this knowledge during inference.

        • All enterprise data: Applies to all Q-Data resources in your enterprise.

        • Apply to specific data only: Select the Q-Data resources to which the rule applies. You can search by Q-Data resource name.

      • In the Application method section, choose whether to Enable forced rewriting.

        Note

        When you enable forced rewriting, if a user's question matches the regular expression, the system rewrites the matched portion to the data interpretation. Use this option with caution.

    3. Click Save.

      To add another entry, click Save and continue adding.

    Batch import

    1. Click Batch import to open the Batch import regular expressions dialog box.

      image

    2. Click Enterprise knowledge base - regular expression template to download the template.

      image

    3. Click to upload or drag and drop a file.image

      Note

      Only .xls, .xlsx, and .csv files are supported. The file size must be 5 MB or less.

    4. After uploading the file, click OK.

      image

    5. After a successful upload, the imported entries appear in the knowledge base.

    Batch export

    Click the image icon to batch export all regular expression match rules.

    image

    Dataset selection

    On the Q-Data knowledge management page, click Dataset selection and configure the following settings.

    Add dataset selection rule

    image

    1. In the upper-right corner of the Q-Data knowledge management - Dataset selection page, click Add dataset selection rule.

    2. Add a business definition, dataset selection, and synonym.

      • Business definition: Define keywords used to match user questions. When a match occurs, the system assigns the specified dataset. Maximum length: 100 characters.

      • Dataset selection: Associate the business definition and its synonym with the datasets used to answer related questions. You can select up to 10 datasets.

      • Synonym: Define alternative names for the business term. This helps the model recognize different ways of asking the same question.

    3. Click Save.

      To add another entry, click Save and continue adding.

    Batch import

    1. Click Batch import to open the Batch import dataset scope page.

      image

    2. Click Enterprise knowledge base - dataset scope template to download the template.

      image

    3. Click to upload or drag and drop a file.image

      Note

      Only .xls, .xlsx, and .csv files are supported. The file size must be 5 MB or less.

    4. After uploading the file, click OK.

      image

    5. After a successful upload, the imported entries appear in the knowledge base.

    Note

    If dataset selection rules are insufficient for complex questions, you can use AI-powered table selection as a supplement.

    Batch export

    Click the image icon to batch export all dataset selection rules.

    image

    AI table selection

    When dataset selection rules are insufficient, a large language model selects the appropriate dataset based on the question's semantics.

    Add AI table selection prompt

    image

    1. On the Q-Data knowledge management - AI-powered table selection page, click Add AI-powered table selection Prompt in the upper-right corner.

    2. Set the effective scope.

      • Select datasets from all queryable data: Apply this prompt to all datasets.

      • Select datasets by analysis topic: Select the analysis topics to which this prompt applies.

        image

    3. Set the prompt.

      Consider the following points when setting the AI-powered table selection prompt:

      • Business metrics and terminology explanations (industry-specific jargon, aliases, and default metrics).

      • Table business context (the facts each table represents, its granularity, its point-in-time or period, and its relationships with other tables).

      • Table selection constraints for typical questions (which questions must use which types of tables, which combinations are allowed, and which mixes are forbidden).

    4. Click Save only or Save and enable.

      • Save only: Saves the current AI-powered table selection prompt.

      • Save and enable: Saves and enables the AI-powered table selection prompt. You can also disable it later from the status column.

        image

    Knowledge base management

    Enable or disable

    The Enabled column shows the status of each knowledge base entry.

    image

    • An entry is enabled if its icon is image. Click the icon to disable it.

    • If the icon for an entry in the Enabled column is image, the entry is disabled. Click the icon to enable it.

    Edit

    To edit an entry, click the image icon in its row.image

    Delete

    To delete an entry, click the image icon in its row.

    image

    Configuration recommendations

    1. For business definitions and synonyms:

      1. Avoid defining short, generic terms such as "take a look", "recent", "generate", "compare", "distribution", and "trend".

      2. Avoid defining common BI terms, such as "time", "maximum", "preview", "data", and "bar chart".

    2. For data interpretations:

      1. Use clear and unambiguous descriptions, as in the following examples:

        1. "Business definition: critical period" - "Data interpretation: March 2020 to March 2021".

        2. "Business definition: business progress" - "Data interpretation: monthly sales and customer count".

      2. Avoid vague time ranges, such as "the following year" or "end of X month".

    3. The knowledge base does not support fuzzy matching or inference. A rule is triggered only when a user's question explicitly contains the defined business term or its synonym.