Strix

Appearance

Related documents ​

The Related documents section in Strix allows users to explore documents that are semantically similar to a selected document. This feature is designed to help users uncover connections, patterns, and relationships across the dataset, enabling deeper analysis and discovery.

Key features ​

  1. Top related documents

    • Strix retrieves a ranked list of related documents based on semantic similarity.
    • Users can view the top 10, 20, 25, or 50 related documents, depending on their selection.
    • Each related document is displayed with key details, including its title, snippet, and similarity score.

  2. Interactive document preview

    • Each related document includes:
      • Title: The title of the document.
      • Snippet: A short excerpt or highlighted text from the document to provide context.
      • Corpus name: The collection to which the document belongs.
      • Token count: The number of tokens (words or word-like units) in the document.
      • Year: The year the document was created (if available).
      • Source link: A clickable link to the document's source (if provided in the metadata).

  3. Graph visualization

    • For certain modes, users can visualize the relationships between the selected document and its related documents using a graph view.
    • The graph displays nodes (documents) and edges (connections), with the size and color of nodes representing their similarity scores and metadata attributes.
    • Users can toggle between the graph view and the document list view for flexibility.

  4. Filtering and pagination

    • Users can filter related documents by metadata attributes such as corpus, year, SweFN, and Blingbring.
    • Pagination controls allow users to navigate through the list of related documents, with options to adjust the number of items displayed per page.

  5. Mobile-friendly design

    • The interface is optimized for mobile devices, ensuring a seamless experience with responsive layouts and collapsible controls.

Example view ​

Below is an example of the Related documents interface, showing the list of semantically similar documents and the graph visualization:

Related Documents in Strix
Related Documents in Strix

Click on the image to view it in full size

How it works ​

  1. Viewing related documents

    • When a user selects a document, Strix retrieves the top related documents based on their semantic similarity.
    • Users can explore these documents in a list view or switch to the graph view for a visual representation of relationships.

  2. Exploring the graph view

    • The graph displays related documents as nodes, with edges representing their connections to the selected document.
    • Users can interact with the graph by zooming in/out or clicking on nodes to view more details.

  3. Filtering and navigation

    • Use the filtering options to refine the list of related documents based on specific metadata attributes.
    • Navigate through the list using pagination controls and adjust the number of items displayed per page.

This feature is designed to provide users with an intuitive and interactive way to explore related content, making it easier to uncover meaningful relationships and gain deeper insights into the dataset.