> For the complete documentation index, see [llms.txt](https://docs.aisera.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.aisera.com/aisera-platform/content-generation-from-tickets/similar-documents-page.md). # Similar Documents Page When your company uses an enterprise application, keeping knowledge documents up-to-date is both critical and time-consuming. With thousands of documents to manage, updating them every quarter (or at regular intervals) is often impractical. It requires significant manual effort to identify new updates, detect outdated content, and determine what information is missing. The **Similar Documents** feature is designed to address this challenge using advanced similarity logic. The Aisera GenAI platform compares all crawled documents from external enterprise application data stores and compares them with AI-generated documents within your Aisera tenant data store. This comparison identifies closely related AI-generated documents for each crawled document and highlights AI-suggested updates by analyzing similarities across content. AI-suggested updates are presented at the section level, making it easier for your reviewers to evaluate and act on relevant changes. The Similar Documents feature allows reviewers to identify outdated documents and review changes, saving time and effort. ### To Compare Documents: Navigate to **Content Generation -> Similar Documents** in the Aisera Admin UI. This opens the **Document Has Similar Knowledge** window.
### Knowledge Clusters The **Knowledge Cluster** section displays crawled documents that are associated with similar AI-generated documents, sorted in descending order (similar to the **Knowledge Generation** page). If you hover a cursor over a cluster in this section, it display a summary with: * **AI Generated Documents**: 2 * **Title of the Document** Selecting a document in the cluster displays only that specific knowledge article. ### Fields in the Similar Documents Window The following table describes fields in the Similar Documents window. | **Field** | **Description** | | ------------------------------------------------ | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Customer-Authored Knowledge Document (clickable) | Displays the knowlege article title. This will be clickable upon clicking, the user will be navigated to the specific knowledge Document details page, where user can review the document with AI suggested content. | | ID (clickable) | Shows the External Document ID. This will be clickable and redirect redirected to the corresponding document details in the external document management system. | | Last Updated | Displays the date when the document was last updated. | | Data Source (Clickable) | Indicates the data source from which the document was ingested into Aisera. Upon clicking it will get routed to the specific Data source page. | | AI-Generated Documents (Clickable) | Displays the count of AI-generated documents identified as similar. This count is clickable. Upon clicking, the user will be redirected to the AI-generated Document Details page, which includes the dropdown having all AI-suggested documents for that specific customer document (as shown in the mockup) for easily to toggle between different documents. | | Review Status | It will show the review status of the document. (Pending Review, Under Review, Reviewed, Reviewed – Helpful, Reviewed – Not Helpful) | ### Filter Options The fields mentioned in the following table will be available as filter selections. (you can search the Document Title along with the ID in the search box, and rest of the fields will be available as filter options). | **Fields** | **Description** | | ------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Last Updated |

Same as SOR -> Tickets last updated, displays the set of duration options:

| | Data Source | Displays the list of Data Sources from which the knowledge article is crawled. | | Review Status |

Displays the drop-down menu values as:

and allows multi-selection.

| ### Document Details Page – Behavior and Content Structure When you select a knowledge article, the **Knowledge Details** window opens. #### Page Layout * The knowledge article is displayed in the left panel. * Each knowledge article section shows corresponding AI-suggested content, if available. * AI-suggested content is expandable and is collapsed by default. #### Section Content Structure Each knowledge article section consists of three layers of content: 1\. **Editable Section Content** (User-Authored) * This is the primary editable content, accessed via the **pencil icon**. * During the Review process, you can: * Edit the selection * Add your own content * Copy the AI-suggested content into the knowledge article * Apply the AI suggested content (over-write it with AI suggested content) * Ignore the AI suggestions * Leave the section unchanged if original content is sufficient. The edited content will not be automatically over-written, removed, or updated throughout the review process. The Aisera GenAI platform always stores the latest user-edited version until the document is marked as: * **Reviewed** * **Reviewed – Helpful** or * **Reviewed – Not Helpful**
2\. **Original Content** (Reference Only) * This is static, read-only content crawled from the external (enterprise application) data sources. * It is available for reference via the **Original Content** button. * This reference helps you compare and validate your changes during **Review**. * Original content is updated when new updates are found in the new crawl and are **approved** and **committed** during a **Review**. In other words, the content that is available in **SOR-> Knowledge** is shown in the original content section. * You can copy this content back into your edited knowledge article and commit it during the Review process.
3\. **AI-Suggested Content** (Read-Only) * AI-suggested content is static and non-editable. * It is updated only when: * New similarities are detected * The similarity job is re-run You may choose to apply or ignore these suggestions. * The **Copy** button allows you to copy the AI-suggested content and paste it into the editable version of the knowledge article section. * If you prefer to use the AI-suggested content instead of the existing content, you can click **Apply**. This will overwrite the editable section with the AI-suggested updates. This is faster than using Copy (above) and then Paste. {% hint style="info" %} **NOTE:** AI-suggested updates display the complete section content, combining the original content with suggested improvements in a clear and easy-to-understand format. This makes it easy to use the suggested content directly. {% endhint %} By default, all AI suggestions are in collapsed mode.
4. **Document Review Status Rules** | **Status** | **Condition** | | ------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Pending Review | None of the sections has edited/ no actions are performed. | | Under Review | At least one section is edited or unlink any of the AI suggested document or updates the section which has no AI suggestions then the status will be changed to ‘Under Review’ automatically. | | Reviewed | When user click on ‘Reviewed’ from Action drop down then the status will be changed to ‘Reviewed’ and user cannot perform any action on the reviewed document which is edit, unlink the AI suggested content etc. | |

Reviewed – Helpful &

Reviewed – Not Helpful

|

Once a document is in the ‘Reviewed’ state, the user has the option to mark it as Helpful or Not Helpful.

Based on this selection, the system will consider the document in the next similarity check. This flag ensures that recent user edits are not immediately overridden, giving teams enough time to review and use the updated content in their source systems.

To support this, we introduced this flag in addition to the review status. Once the document is marked as Helpful or Not Helpful, it will be included in the next run, where the content may be updated with the latest AI-suggested changes along with the original crawled content.

| ### Helpful / Not Helpful Flag The **Helpful** / **Not Helpful** flag gives you control over when reviewed documents should be considered for further processing, without automatically updating your source systems. When a document is in the **Review** state, you will see two options: * **Helpful** * **Not Helpful** You can select only one option. * Once selected, the choice cannot be changed. * A confirmation message will be shown before applying the selection. #### The Helpful/Not Helpful Flag: * Prevents automatic updates to your source system before your internal approval is complete. * Ensures your existing approval workflows remain unchanged. * Gives teams time to review and adopt changes at their own pace. **Not Helpful Means:** * The suggested content is not suitable for use. * The knowledge article will still be considered in future runs, allowing improved suggestions. **Status Behavior** * Until a document is marked **Helpful** or **Not Helpful**: * It remains in the **Reviewed** state. * It is not included in further similarity checks. * Once marked: * The document becomes eligible for the next similarity/deduplication run. * The system will use the latest version available from your source system. #### How It Works 1. A document is Reviewed in the Aisera Gen AI platform. 2. The Reviewer marks it as **Helpful** or **Not Helpful**. 3. If marked **Helpful**, you can copy the updated content and apply it back to your enterprise applicat data source (original source) as part of your company's approval process. 4. Once approved in your system, Aisera Gen AI platform will crawl the latest version back. 5. The updated document is then available for downstream use. #### Important Notes * During the Reviewed state, updated content is stored only within Aisera and is not pushed to your source system. * If you mark a document as Helpful, make sure to copy the content to your source system before the next run. * In the next run, the system will use the version from your source system and may refresh the document with new AI suggestions. #### Confirmation Messages * **Helpful**: \ Thank you! Please copy the updated content to your source system. Otherwise, in the next run, the reviewed changes may be lost. * **Not Helpful:** \ Thanks for your feedback. We’ll provide improved suggestions in the next run.
### Unlinking Documents During the review process, if you determine the AI-suggested knowledge article is not similar or relevant, you can unlink it. This captures feedback and improves the similarity algorithm. The **Unlink** option allows you to remove the association between documents when they are not relevant. #### How Unlinking Works * When you click **Unlink**, a confirmation popup appears, asking you to confirm the action. Upon clicking **Yes**, the corresponding AI-generated knowledge article is greyed out, indicating that it is no longer considered similar. This also helps improve the accuracy of future similarity checks from a data science perspective. * If you select Cancel, no changes are made and the knowledge article remains linked. **Impact of Unlinking** * If the AI-suggested knowledge article was linked to only one original knowledge article: * It will be removed from the **Document Has Similar Knowledge** page. * If it was linked to multiple documents, the count will be updated accordingly (for example, from 3 to 2). ### AI-Suggested Documents * AI-suggested knowledge articles in the right-side panel are shown in a collapsed view by default. * When expanded, you can view the **Document Title**, **Generated Date**, and **Similarity Score**, as shown in the UI. * Citation counts are displayed at the section level (such as, “3 citations” next to a section like **Resolution**). * You can open the complete document in a new tab using the **Open Full AI-Generated Document** link. #### Within the Full Document View **You can:** * View all AI-suggested knowledge articles * Switch between them using a drop-down menu * **Edit** knowledge article content * **Publish** documents * Provide **Feedback** ### Citations AI-suggested content is a consolidated summary derived from up to five AI-generated knowledge articles. Each section includes all relevant citations. Selecting a **Citation** expands the corresponding section in the AI-generated knowledge article for easy comparison. If an AI-suggested knowledge article is **Unlinked**, its associated **Citations** will be removed or disabled in the suggested content. ### Handling Deleted AI-Generated Documents In some cases, an AI-generated document may be deleted during the review process. To avoid confusion, such documents will still appear in the side panel, but they will be greyed out and no actions can be performed on them. This ensures clarity, especially in cases where certain citations in the suggested content may have originated from those deleted documents. ### How AI-Suggested Content is Determined To provide accurate and relevant AI suggestions, the system follows specific rules when selecting crawled documents and AI-generated documents for similarity checks. #### Crawled Documents (Source Content) **Rule 1:** Section-Based Content Processing * AI suggestions are generated at the section level. * During document ingestion, headings and subheadings from your source system define how content is structured. * To ensure accurate results: * Use proper heading styles in your source documents. * The system relies on these headings to identify and segment sections correctly. * Only the primary (base) headings are considered for section identification. **Rule 2:** Active Documents Only * Only active crawled documents that are approved and commited are included in similarity checks. In other words the documents that are in SOR-> Knowledge page are considered for similarity check. * Inactive documents are excluded to maintain relevance. **Rule 3:** Source of Truth for Base Documents * Only customer-owned documents ingested from source systems are treated as base documents. * AI-generated documents published within the Aisera platform are not considered base documents. * However, if an AI-generated document is: * Exported to your source system, and Crawled back into the platform, \ it will then be treated as a base document. **Rule 4:** Only the base documents that are ingested from the Data sources that are attached to the application are considered. ### AI-Generated Documents **Rule 1:** Only active AI-generated documents are considered during similarity checks. **Rule 2:** Application-Specific Scope * AI-generated documents are evaluated within the context of the specific application they belong to. * Only documents generated for the same application are included in the similarity analysis. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.aisera.com/aisera-platform/content-generation-from-tickets/similar-documents-page.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.