The content landscape can be accessed through the “Explore” view. It allows you to visually discover data and identify patterns and trends. After clicking the “Explore” icon, wait for the landscape to generate.

Once generated, the landscape will display labels, providing a visual representation of the content clusters from the Knowledge Base(s). This helps you explore and gain an overview of the data.

Begin by navigating the overall landscape to identify key areas of interest. Larger, more defined clusters circled in yellow indicate prominent themes within the dataset.

As you zoom in, the number of labels on the hills increases, offering a more detailed view of the information. Zooming all the way in reveals the individual documents that comprise your landscape. In other words, zooming in brings you closer to individual documents, while zooming out provides a bird's-eye view.

Red areas: These areas indicate a high concentration of semantically related documents or discussion topics. The deeper the red, the more closely related the topics are. Red areas consist of numerous closely connected topics or documents. If a landscape contains many red areas, it signifies a high level of diversity within the dataset.

Blue areas: These areas represent regions with less related content. Lighter shades of blue indicate areas with fewer connections or discussions around a particular topic.

Clicking on the black labels in the landscape reveals a summary of the texts in that area, along with references to the original texts.

Yellow dots that appear when you zoom in reveal individual documents. Clicking on a yellow dot displays the original document content along with its metadata.

Pipeline sidebar opens and closes through the little arrow on the right side of the page. It displays various settings available in the Insight Booster, allowing you to tailor your analysis with different filters and configurations.

When certain landscape parts are irrelevant, or if you want to explore a specific area in more detail, filters are useful.

You can filter the landscape using contour selection or any of the metadata fields from the Knowledge Base. To select a contour, click on a layer in the landscape. Filters can be applied locally (“Filter within the landscape”) or globally, which will regenerate the landscape (“Filter and regenerate the landscape”). You have the option to filter in (“Focus on selected”) or filter out (“Ignore selected”).

Remember that regeneration filters take longer to load as the entire landscape is re-computed.

Size field: If you have any numeric metadata, you can use it to determine bubble size.

Color field: Select categorical metadata fields to color the documents.

Semantic outlier removal policy: When the landscape is generated, small clusters at the periphery are automatically removed by default. You can change this setting to “Always” or “Never” to adjust the removal criteria. If you want all documents displayed, we recommend selecting the “Never” setting.

Language: By default, topic summaries and titles are generated in English, regardless of the language of the input document. You can select a different language here for the generated content.

The action menu found at the top left part of the landscape provides quick access to key functions for interacting with the data:

Icon 1 - Switch to feed: Exists the landscape and creates a feed to dive into the individual documents. Through the feed mode, you can change how documents are displayed (stream, long, or short), sort them in your preferred order, and perform a semantic search to find specific documents. Using the “Switch to landscape” button, you can go back to the content landscape.

Icon 2 - Download: Export documents in SVG, PNG, or Excel formats. The Excel output includes topic names and metadata.

Icon 3 - Publish/copy link: Publish a public URL to share your results externally. You can unpublish using the same icon. Ensure public sharing is enabled through the list of Insight Booster projects.

Icon 4 - Date coverage: See the time period covered in your data, if dates are available.

Icon 5 - Total number of documents: See the total number of documents displayed in the landscape. This number will be updated with the applied filters.

Icon 6 - Search: Perform a semantic-based search across the landscape to refine your data exploration.

The Research Chatbot is designed to assist you by utilizing the Knowledge Bases specific to your project. To begin interacting with the Research Chatbot, simply click on the "Research Chatbot" tab located at the bottom of the page.

Clicking the 'Question-answer' button will open a menu that allows you to switch between different interaction modes. The Research Chatbot supports two modes:

Question-answer: This mode functions like a chatbot, allowing you to ask questions and receive answers sourced directly from your knowledge base. The primary advantage is that its responses are tailored specifically to your data.

Subtopic summarization: This mode also provides answers to your questions, but it specifically focuses on summarizing themes or events into subtopics.

You can type in questions, and the answers you receive will be based on the documents in your knowledge base, complete with downloadable references. Once you ask a question and receive answers, the relevant area in the landscape will light up, allowing you to view the documents connected to your question.

By clicking the downward arrow in the top right corner of the response, you can view or download references as an Excel file.

If you click on “Show references”, you will see the references sorted in the order they appear in the response, along with links to the original sources (if available).

To delete your chat history, click the trash icon.