> ## Documentation Index
> Fetch the complete documentation index at: https://docs.converlens.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Clusters (Topics)

# About Clusters

Clusters help visualise the relationships between different topics or themes that emerge from your dataset. Each cluster represents a distinct topic or theme, with the proximity of clusters indicating their similarity or relatedness.

The size of each circle reflects the importance or prevalence of a topic within the dataset.

* Larger circles represent highly frequent or dominant topics.
* Smaller circles indicate less frequently mentioned topics.

**Tip:** Clusters are a great way to understand the connections between the words you encounter in the Words & Terms section and the related topics discussed in responses. Especially with extensive datasets, these clusters provide a valuable starting point for exploring responses and generating ideas for structuring your tagging system. They also help pinpoint areas for deeper analysis through tags and highlights.

# Where to find Clusters

To access Clusters, navigate to the Explore tab and select “Clusters” from the Navigation Tabs:

# Clusters options

## Cluster Level (Per Record or Per Sentence)

Toggle between:

* **Per Record:** Groups topics based on entire responses.
* **Per Sentence:** Identifies themes within individual sentences, providing more granular insights. (Recommended for capturing nuanced topics)

## **Number of clusters**

Select between 3 to 10 clusters, depending on how broadly or narrowly you want to group topics.

## View as plot or topic distribution

Choose how to visualise the clusters:

* **Cluster Plot:** Displays a spatial representation, showing the relationships between topics.
* **Topic Distribution:** Displays a ranked list of topics as a bar chart based on prevalence in the dataset.

## **Set Topic Level**

* **Primary Topic Only:** Focuses on the main topic for each response.
* **Include Secondary Topic:** Display a second topic ring for each cluster.

# **Understanding Cluster Topics**

## **Topics & Representative Words**

Each cluster includes:

* **Topics:** The broad topics identified in responses.
* **Percentage**: How frequently each topic appears in the dataset.
* **Representative Words**: Key terms within each topic, ranked from low to high saturation, with a percentage indicating their prominence.

# Matching records

Displays the individual responses that match the currently selected topic.

* Representative words in each response are highlighted for easy identification.
* Click any response to open it in Submission Reader for further review.

## Selecting and Tagging Responses

To tag responses at the record level:

1. Select responses using the checkboxes.
2. Click the Tag dropdown.
3. Choose a tag and click Add.
4. Confirm by selecting OK—the tag will be applied to selected responses.

# **Using the Cluster Tool**

## **Analyse Button**

Once settings are configured, click Analyse to process your dataset. The tool will:

* Read through responses.
* Group sentences or responses discussing similar or related topics.

## **Topic Selection**

Click on a topic to:

* View its representative words.
* See matching responses related to that topic.

### **Refresh Button**

Click Refresh to re-run the clustering process.

**Note:** Cluster results may vary slightly with each refresh, as clustering algorithms dynamically adjust based on the dataset.

### **Download Cluster Chart**

<img src="https://mintcdn.com/converlens-654e536e/56mimSbn1jWqueAp/docs/images/01-03-04-05-07-01__clusters-topics-01.webp?fit=max&auto=format&n=56mimSbn1jWqueAp&q=85&s=458bd0dc24e3ed4ddae463e5aa357093" alt="Clusters (Topics) screenshot: Download Cluster Chart" width="1096" height="710" data-path="docs/images/01-03-04-05-07-01__clusters-topics-01.webp" />

If you want to save your cluster visualisation, click the Download button to export the chart.